Technical and Structural Effects of China’s TFP Growth
Abstract: TFP growth may derive from both technology progress (technical effect) and factor allocation (structural effect). Using China’s macroeconomic and industrial data, this paper decomposes China’s TFP growth on the basis of growth accounting to cast light on China’s growth sources since reform and opening up in 1978. Our study has led to the following findings: (1) From 1978 to 2014, China’s economic growth was of generally good quality, and about 1/3 of growth momentum stemmed from a general technology improvement. (2) After 2005, China’s late-mover advantage diminished due to narrowed technology gaps with advanced economies. This resulted in a sharp decline in the contribution of technology progress to growth. However, structural effect contributed a steadily increasing share to China’s growth. (3) After global financial crisis in 2008, there has been a tendency of reverse technology progress in terms of factor allocation in sectors with excess industrial capacity and other sectors like finance and real estate. Therefore, China should divert its factor resources to more tech-intensive and efficient sectors in the short run, and strive to promote technology progress in all sectors in a longer timeframe.
Keywords: TFP, technical effect, structure effect, growth accounting JEL Classification Codes: O47; O33; O14
DOI: 1 0.19602/j .chinaeconomist.2018.09.04
Since 2014, China’s economy has entered into the new normal with increasing downward pressures. Given the urgency to boost growth, CPC Central Committee General Secretary Xi Jinping called for “enhancing China’s overall productivity through supply-side structural reforms” in November 2015. Later, he noted that “supply-side structural reforms must improve total factor productivity by reducing
1 ineffective supply and increasing effective supply.” These guidelines are reaffirmed in the Outline of China’s 13th Five-Year Plan (2016-2020).
In his remarks, General Secretary Xi Jinping stressed the importance of supply-side structural reforms to total factor productivity (TFP) and growth sustainability. This argument is supported by growth economic theories. Total factor productivity (TFP) determines the level of output that can be achieved with a given combination of factor inputs. TFP variations, i.e. “TFP growth” or “TFP index,” are more often discussed in economic analysis. At the micro level, we may estimate corporate TFP growth from a production frontier perspective, and decompose it into two parts, including “movement on the production frontier” and “change in the level of corporate technology relative to production frontier.” At the macro level, in addition to technology progress, TFP growth may also be realized by allocating production factors to more productive industrial sectors as well.
These two methods can be referred to as the technology progress effect (“technical effect”) and industrial restructuring effect (“structural effect”). The latter is how overall TFP and growth potentials can be achieved through supply-side structural reforms. Estimating both effects helps obtain information about technology progress, inter-industry factor flow and allocation efficiency. Such information can be used by the government in guiding resource allocation. In existing measurement practices, TFP growth that appears in the form of residual value is most often equated to technology progress, and rarely further decomposed.
Based on Jorgenson growth accounting framework, Divisia index, Massel (1961) and the shift-share method, this paper includes industrial sector TFP index to mathematically decompose the expression of overall TFP growth. According to the decomposition model, we divide China’s overall TFP growth sources since reform and opening up in 1978 into technical effect and structural effect on the basis of estimating China’s overall and sector-specific TFP growth, so as to identify China’s growth momentum and propose suggestions on improving China’s overall TFP and growth potentials.
2. Literature Review and Research Methodology 2.1 TFP Growth Estimation and Macroeconomic Growth Accounting
TFP aims to measure the economic efficiency. In a single- input- single- output ( SISO) system, TFP can be expressed as “input-output ratio.” If labor is the sole input, the result is labor productivity, which is a common concept in economics. In reality, however, labor productivity cannot reflect overall productivity. It is thus necessary to measure the output efficiency of a combination of all observable factor inputs. This need is satisfied by TFP and TFP growth estimation (Hulton, 2000; Syverson, 2011). Depending on the object of assessment, TFP growth estimation can be carried out for the corporate sector, an economy or its individual industrial sectors.
Estimation of corporate TFP growth is based on the relative efficiency approach with production frontier as benchmark, and originates from Farrell’s (1957, 1962) groundbreaking work. Considering general manufacturers’ “multiple- output multiple- input” characteristic, Farrell ( 1957) uses equalproduct curve ( production frontier) to measure manufacturers’ input- output efficiency. Production frontier represents the highest level of technology. An input-output combination on the production frontier is technically the most efficient: The closer it is to the equal-product curve, the more technically efficient it is. Relative efficiency with production frontier as benchmark can be converted into distance function in mathematics, which is the basis for estimating TFP index (Malmquist, 1953; Shephard, 1953, 1970). According to the difference of distance function expression, frontier TFP index estimation can be divided into data envelopment analysis (DEA) and stochastic frontier analysis (SFA). Using mathematical planning, DEA converts the distance function estimation of relative efficiency into the solution of linear planning objective function, and combines technical efficiency (distance function) with Malmquist index to estimate TFP indexes between different time points. This method is also referred to as non-parametric method, since it does not involve the specific form and parametric estimation of
production function (Charnes & Cooper, 1962; Charnes et al., 1978; Banker et al., 1984). SFA depicts manufacturers’ production behaviors through stochastic production frontier function. The stochastic error term of production function is divided into the symmetrical error term of the effect of various stochastic environmental factors on frontier production and one-side error term that measures technical inefficiency, i.e. manufacturers’ technical efficiency (distance function). The combination of technical efficiency (distance function) and Malmquist index thus obtained may also be used to estimate TFP index (Aigner et al., 1977; Meeusen & Broeck, 1977).
Stigler, Abramovitz, Solow, et al. all made groundbreaking contributions to TFP growth estimation. Among them, Solow neoclassical growth model and the famous “Solow residual” and macroeconomic growth accounting system are the most influential (Abramovitz, 1956; Solow, 1957). Solow model decomposes growth sources into the three parts of capital, labor and “neglected factor (Solow residual),” for estimating the contribution rates of different factors. Growth of the “neglected factor” is TFP growth. After Solow, Jorgenson and Griliches introduced investment theory, index theory, national income accounting system and corporate theory into growth accounting framework, thus developing a complete and stringent growth accounting framework. Considering Jorgenson’s outstanding contribution, we refer to it as “Jorgenson growth accounting framework,” which integrates growth source decomposition with national accounting system. With respect to capital input estimation, we include such concepts as capital services, diminishing production capacity, and retirement and service life of inventory capital (maximum length of service). With respect to labor, we take into account labor quality aspects like education and health. This more reasonable system has been extensively applied globally. In order to standardize TFP growth estimation and increase the comparability of results, OECD made detailed explanations on Jorgenson growth accounting framework and TFP growth accounting (OECD, 2001).
In China, overall TFP growth estimation can be traced back to at least the early 1990s. A CASS team led by Li Jingwen conducted a comparative study on productivity in China, the US and Japan in collaboration with Jorgenson, Masahiro Kuroda, et al. (Li, et al., 1993; Li and Li, 1993; Zheng, 1998). In early 21st century, overall TFP received more attention, as evidenced in the relevant studies of Huang, et al. (2002), Sun and Ren (2005), and Guo and Jia (2005). In recent years, Chinese scholars also attempted to estimate the overall or sector- specific TFP index using production frontier method. Generally, enterprises or regions are regarded as efficiency and TFP index estimation units, and weighted average sector-specific or overall TFP index is calculated on the basis of measuring the efficiency and TFP index of each unit (Wang Zhigang, et al., 2006; Yao Zhanqi, 2009).
2.2 Industrial Restructuring and Decomposition of Overall Productivity Index
Apart from the level of technology, industrial restructuring is also an important factor that influences TFP. According to Petty-Clark Theorem, in the economic takeoff stage, primary industry will reduce in proportion, and secondary and tertiary industries will increase, resulting in a dominant share of secondary industry. With economic sophistication, primary and secondary industries will decline in proportion, while tertiary sector expands into a new, dominant industry (Clark, 1940). Growing share of secondary industry, which is much more productive than primary industry, will unleash an economy’s potential productivity. After industrialization completes, factor resources keep moving to secondary and tertiary industries. However, stagnant productivity in many tertiary sectors leads to a reduction in overall productivity (Baumol, 1967; Kruger, 2008). By creating a multi-sector growth model, Montobbio (2002), and Ngai and Pissarides (2007) investigate dynamic structural changes in the growth process, and redepict the pattern of above-mentioned industrial structural evolution and its impact on productivity.
Empirically, Dietrich (2009) utilizes the panel data of seven OECD countries and tools such as panel Granger causality test to examine the relationship between industrial structural change and economic growth. However, the relationship of causality shown by the empirical study is uncertain. More empirical studies divide variations in overall productivity into overall technology progress and industrial
structural change in order to examine the effect of industrial structural adjustment on productivity. Peneder (2003), Fagerberg (2000) et al. adopt deviation-share decomposition method to decompose labor productivity index into industrial progress effect and industrial structural change effect2. After decomposing the data of 28 OECD countries, Peneder (2003) discovers that industrial progress is the decisive factor of productivity improvement, and that the productivity effect of structural change can be positive or negative but is limited. Fagerberg (2000) focuses on the manufacturing productivity effect of specialization and structural change. His empirical study on 24 sectors in 39 countries shows that although on average structural change does not have any significantly positive effect on productivity, countries experience faster productivity growth if the share of sectors with rapid technology progress increases.
Labor productivity improvement results from both technology progress and capital deepening. The industrial progress effect, which is expressed by various sectors’ weighted labor productivity index, is not totally equal to technology progress effect. This problem can be solved by similar decomposition of TFP index. Massell (1961) extends Solow model and growth accounting to industrial sectors, and decomposes overall TFP growth into weighted TFP growth and structural changes stemming from intersector flow of capital and labor. While the former measures the technical effect, the latter measures the structural effect of structural transition. Massell (1961) divides the US economy in the 1950s into 19 sectors, and the result shows that technical effect contributes about 2/3 of US TFP growth while structural effect contributes the rest 1/3.
Chinese scholars also carried out empirical studies from an index decomposition perspective. Based on data of China’s primary, secondary and tertiary industries, Li (2011) decomposes TFP growth into weighted TFP growth of various sectors, inter-sectoral labor flow and capital flow, which leads to similar conclusions with Massell (1961). Wang et al. (2004) conducts an empirical study on the relationship between structural adjustment and productivity based on Solow growth model, but employs econometric regression based on micro-level corporate data. Using frontier approach, Yao (2009) estimates sectoral TFP index, and investigates the effect of factor allocation on TFP growth.
2.3 Review of Existing Studies and Our Approach
Since the 1950s, the academia has gradually formed a relatively sophisticated system of methodologies for estimating TFP index. Estimation of TFP index at the micro level mainly employs DEA and SFA, and overall TFP index estimation relies on growth accounting framework. Micro TFP index can be decomposed into technical change index and efficiency change index, both of which reflect the effect of technology factor. 3Aside from overall technology progress, industrial structural change will also bring about change in overall productivity. Hence, overall productivity index can be decomposed into technical effect and structural effect. For convenience, most relevant overseas empirical studies chose to decompose labor productivity index, and very few studies decomposed overall TFP index. Chinese scholars carried out extensive empirical research on growth accounting and TFP index estimation. However, such details as capital input and labor input estimation in some studies need to be further refined. As for the relationship between shift of industrial structure and productivity growth, many Chinese studies were carried out to measure micro-level corporate and industry TFP indexes using methods like DEA and SFA, and few empirical studies were carried out using growth accounting
and overall TFP index estimation. In addition, existing Chinese studies made a rough classification of industrial sectors in analyzing the structural effect.
In order to more clearly and precisely estimate and decompose the technical effect and structural effect in China’s overall TFP index, this paper will collect overall data and data from 17 sectors to carry out an empirical analysis based on the creation of a decomposition model framework for overall TFP growth estimation. The rest of this paper is arranged as follows: Part 3 describes the decomposition model for estimating overall TFP index; Part 4 explains data treatment and shows key results of estimation and decomposition; Part 5 offers an in-depth analysis of the result of decomposition analysis; and Part 6 is concluding remarks and policy recommendations.
3. TFP Index Estimation, Decomposition and Factor Estimation Model 3.1 TFP Index Estimation Model
According to the definition, TFP is a ratio of total output combination relative to total input combination. Thus, we have: In equation (1), A represents TFP, Y is output and X is input. We use A, and to respectively denote the differentials of TFP, output and input with respect to time, and take logarithm of both sides of equation (1), which gives us: Assuming that return to scale is constant in the production function and return to factor is equal to its marginal output, we may obtain based on Divisia index: and in equation (3) respectively denote the share of various output and factor inputs in the total value, and meet: = =1. ≥0, ≥ 0.With capital and labor as the only two factor inputs taken into account, equation (3) can be simplified and extended into various sectors: in equations (4) and (5) denotes labor output elasticity, or the share of labor input in total input (value).
3.2 TFP Index Decomposition Model
Capital input and labor input as a share in total input of various industrial sectors are specified as and respectively, and change in the factor input of various industrial sectors and total output growth can be expressed as:
Where, is the weighted value of technology progress for various sectors, which roughly reflects changing technology and denotes the technical effect of overall TFP growth. and respectively reflect the flow of capital and labor across various sectors, i.e. structural change in factor allocation. Positive values of , suggest that a greater share of factors is allocated to sectors with higher marginal output (or higher efficiency), and negative values of , suggest that a greater share of factors is allocated to sectors with lower marginal output4. We refer to and and as the structural effect of capital and the structural effect of labor. + denotes the overall structural effect of overall TFP growth, and is the result of capital and labor reallocation. Equation (10) may also be simplified as equation (14).
3.3 Factor Input Estimation Model
This paper will estimate factor input under Jorgenson growth accounting framework. Capital participates in production in the form of capital services. In estimation, we should take into account fixed capital formation and productivity change (reduction) and retirement (decommissioning). According to OECD (2009), double curve time-efficiency model is employed to depict change in capital productivity, and lognormal distribution depicts its decommissioning mode.
In equations (15) and (16), T is the capital’s (maximum) length of service (or retirement age), n its current year, and parameter b ≤ 1 determines the shape of function. and are the standard deviation and mean value of lognormal distribution function, , . Where, m and s are the mean value and standard deviation of normal distribution behind lognormal distribution function. m is capital’s average length of service, and the scope of ’s value is normally . Greater value means steeper distribution.
Subsequently, perpetual inventory method is used to estimate productive capital stock of type i inventory capital at time point t. The price of such capital (or user cost) is determined, and the two are multiplied to arrive at (the value of) capital input; Where, and are productive capital stock and user cost for type t capital during period i; and are time-efficiency model and retirement model for type i capital. In equation (17), is the investment spending for type i capital during period t, i.e. “fixed capital formation”; is price index. In equation (18), subscript s is the actual length of capital service, q is asset price, r is return to capital,
d is asset depreciation rate, and is change in asset price. This equation also reflects the relationship of user cost conversion for the same type of capital during different periods.
Labor is categorized according to such characteristics as the length of education, and the quantity of input is measured by the unit of labor hour. Different types of labor input can be aggregated using their share in total labor compensation as weight. Hence, labor input growth can be expressed as:
L is total labor input; Li is different types of labor input manifested in the number of labor hours; pi is the price of type i labor input, such as hourly wage; is the share of type i labor compensation.
4. Data Treatment, Estimation and Decomposition Results 4.1 Estimation of Factor Input
After estimating overall capital factor input using a top-down approach, we obtain 17 sectors from reasonable decomposition. Given the differences in productivity change, length of service and retirement mode, inventory capital is divided into the three categories: (1) buildings, (2) machinery and equipment, as well as (3) others for estimation in accordance with the following six steps: (1) collect and arrange fixed capital formation data sequence; (2) select appropriate price index to convert “fixed capital formation” data into comparable price; (3) specify the service length-efficiency model for various types of capital according to the diminishing productivity characteristic; (4) set relevant decommissioning mode; (5) perpetual inventory method is employed to estimate the productive capital stock for various types of capital in each year, i.e. quantity of capital services; (6) calculate user cost for various types of capital in each year, i.e. price of capital service.
Based on Historic Information for China’s GDP Accounting: 1952-2004, Input-Output Table and other statistical information, data of missing years is completed to form a data sequence for the total amount of fixed capital formation during 1952-2014, which is decomposed into data sequences for the three types of fixed capital formation. The investment price indexes of the three types of fixed capital since 1990 are obtained from China Statistical Yearbook to estimate the missing data based on the total value of current-price fixed capital formation and constant-price fixed capital formation growth. Parameters of three types of inventory capital are specified as 0.75, 0.5 and 0.6, which correspond to depreciation lengths of 38, 16 and 20 years respectively. The value of parameter in the decommissioning mode equation (15) is specified to be half of the length of capital service. The value of s is m/ 2 (OECD, 2009; Cai and Zhang, 2015). Now, the inventory of three types of productive capital during 1977-2014 can be estimated5. According to the equivalent relation that the sum between labor compensation and capital compensation equals total output, labor compensation data can be used to obtain (average) return to capita rt for each year. By substituting return to capital into equation (18), we may calculate the user cost of various types of productive inventory capital, and obtain the value of various capital services by multiplying the productive capital stock of various types of capital with respective user cost, i.e. various types of capital input. The result is shown in Table 1.
According to the approach and method of Cai and Zhang (2015), labor input aggregate is estimated with the unit of labor hour, and fully takes into account the distribution of workforce education. Data for estimating the value (labor input aggregate) of various types of labor (time) is extended to 2014. Result
is shown in Table 2.
In order to maintain consistency with existing statistical accounting system, China’s economy is divided into 17 sectors based on data availability (see Table 3 for details). Capital for various sectors is also divided into the aforesaid three types. Their efficiency reduction mode, service length and decommissioning mode are specified referencing OECD (2009) to estimate various types of productive capital stock in these sectors. Based on the estimation result of user cost, we may obtain the value of sector-specific capital services, i.e. sector-specific capital input. Labor estimation is also divided into 17 sectors, and labor input data is estimated for various sectors with labor hour as the basic unit by such criteria as the distribution of educational level.
4.2 Estimation of TFP Growth and Contribution
Using estimated factor input, we decompose the growth of China’s economy and 17 sectors during 1978-2014, with results shown in Table 3 and Table 4.
4.3 Decomposition of Overall TFP Growth
Based on equations (6)-(13), as well as the estimated sector-specific TFP growth and sector-specific factor input, we decompose overall TFP growth during 1978-2014 by technical effect and structural 6 effect for 17 sectors.
5. Analysis of TFP Growth and Its Decomposition Result 5.1 Contribution of TFP Growth to Economic Growth
Based on overall and sector-specific TFP contribution to growth in different stages shown in Table 3 and Table 4, the following assessment can be made.
( 1) TFP played a very important supportive role to China’s rapid growth with an average contribution of 39.4%. Before 2000, the contribution of TFP growth experienced significant volatility. After 2005, there was a clear downward tendency.
(2) Agricultural TFP growth played a dominant role in supporting its value-added growth, with an average contribution of 82.2%. Before 2000, there was also significant volatility in TFP contribution. After 2000, TFP contribution steadily increased, exceeding 150% in all years.
(3) TFP growth of secondary industry played an important role to the industry’s value-added growth, with an average contribution of 33.9%. But the contribution had a significant tendency to decline. After 2010, TFP’s contribution to the growth of seven sectors including “food processing sector” was even negative. During the same period of time, capital input contribution increased from 39.2% during 19901995 to 91.8% during 2010-2014.
The above estimation result reflects an enhanced factor-driven and investment-driven characteristic. (4) TFP growth of tertiary industry also played an equally important supportive role in its valueadded growth, with an average contribution of 35.8%, and was generally stable in various stages. After 2000, in particular, the average contribution stabilized at 20% to 40% in various stages.
5.2 Technical Effect and Structural Effect of Overall TFP Growth
On the basis of overall TFP decomposition, we may further estimate the contribution of technical effect and structural effect to TFP growth, and decompose such contribution to primary, secondary and tertiary industries, with some results shown in Table 6. Hence, we may reach the following conclusion:
(1) Since reform and opening-up, technology progress in various sectors has been the primary source of overall TFP growth. During 1978-2014, the average contribution of technology effect stood at 83.7%, and structural effect was only 16.3%. In most periods of time, the technical effects of primary, secondary and tertiary industries were significantly positive. Primary industry exhibited a significant negative contribution after the 1980s. Secondary industry made a positive contribution in most periods of time. Tertiary industry demonstrated a significant and steady positive contribution.
(2) At the beginning of reform and opening up, structural effect contributed significantly to overall TFP growth. In particular, the structural effect of labor contributed a particularly significant share to primary industry, while the contribution of technical effect mainly stemmed from secondary and tertiary industries. A possible reason is that China’s rural reform greatly increased farmers’ enthusiasm.
7 Labor hours of each farmer significantly increased and turned into positive structural effect of labor. By introducing advanced foreign technology and managerial experience through opening up, China vigorously promoted industrial and service sector development.
(3) After 1985, technical effect had an absolutely dominant contribution to overall TFP growth, which lasted after China’s WTO accession in 2001. After the mid-1980s, China’s late-mover advantage was brought into full play; such an advantage derived from the significant technology gaps between China and Western countries. In this period, technical effect made positive contributions to primary, secondary and tertiary industries. The contribution of structural effect had always been negative for primary industry, generally positive for secondary industry, and completely positive for tertiary industry. This indicates the concentration of factor resources in secondary and tertiary industries. This is consistent with the reality of China’s rapid industrialization and the rapid development of secondary and tertiary industries.
(4) After 2000, the contribution of technical effect to overall TFP growth significantly declined, while structural effect greatly increased. These changes largely stemmed from secondary industry. The contributions of technical and structural effects to tertiary industry were relatively stable. A possible reason is that after about two decades of opening up, China greatly narrowed its technology gaps with advanced economies. Technology progress through importation and diffusion became increasingly difficult. But continuous industrialization and urbanization resulted in the allocation of factor resources to more productive secondary and tertiary sectors, which maintained fairly high TFP growth.
(5) After the global financial crisis in 2008, the contribution of technical effect to overall TFP growth steeply declined, and structural effect became a dominant contributor to overall TFP growth. During 2010-2014, the average contribution of structural effect to TFP growth stood at 59.3%. The average contributions of technical and structural effects to secondary industry were -29.4% and 88.2% respectively. This implies that China’s regulatory measures including the four-trillion-yuan stimulus package did not bring about any improvement in the level of technology but instead caused factors to rapidly aggregate in secondary industry. This trend appears to have improved after 2014.
5.3 Technical and Structural Effects of TFP Growth in Secondary Industry
TFP growth of China’s secondary industry is decomposed to calculate the contributions of technical and structural effects, and the contributions of the effects are further decomposed into 11 sectors of secondary industry (see Table 7). In relation to Table 4, the following assessment can be made.
(1) Since reform and opening up in 1978, the technical effect has served as a dominant factor that supported TFP growth in China’s secondary industry. On average, 1/3 of growth in secondary industry stemmed from technology progress in various sectors, while structural effect was generally insignificant.
(2) Before 2005, almost half of the value-added growth of secondary industry stemmed from a general technology progress in various sectors. After 2005, the supportive role of technology progress diminished. After 2010, the overall level of technology in various sectors even reduced, resulting in a negative effect on the growth of secondary industry. During the same period of time, structural effect played a significantly positive role in the growth of China’s secondary industry.
(3) After 2010, technology progress in seven sectors including “food processing” had a negative effect on the growth of secondary industry. Capital and labor inputs contributed 91.8% and 22.4% respectively to secondary industry. This means that what supported secondary industry’s growth after the eruption of global financial crisis is investment and factor expansion, which is especially prominent in
8 sectors with excess capacity. This also started to improve in 2014.
5.4 Technical and Structural Effects of TFP Growth in Tertiary Industry
TFP growth of tertiary industry is decomposed to calculate the contributions of technical and structural effects. The result is fully decomposed into five sectors, as shown in Table 8. Based on Table 4, the following assessment can be made:
(1) Since reform and opening up in 1978, technical effect played a dominant role in supporting the TFP growth of China’s tertiary industry. On average, technology progress in various sectors contributed 1/3 of growth in tertiary sector, but the contribution of structural effect is limited.
( 2) Before 2005, technology effect almost contributed all the TFP growth of China’s tertiary industry. After 2005, the contribution of structural effect significantly increased, reaching 60.3% during 2010-2014, and replaced technical effect as a key driver of TFP growth.
(3) For the five sectors, the contribution of technology progress to tertiary industry mainly derived from other services, including commerce and catering, transportation, warehousing and postal services, while the technical effect of finance and insurance, and real estate is negligible.
(4) The structural effect of finance and insurance, and real estate played a significantly positive role in the growth of China’s tertiary industry. These two sections have attracted a large number of factors despite the stagnation and even retrogression of technology. However, the structural effect of other services, including transportation, warehousing and postal services with the most rapid technology progress contributed negatively to tertiary industry’s growth on an average basis. This implies a significant reverse technology progress tendency in the resource allocation and agglomeration within tertiary industry.
6. Concluding Remarks and Policy Recommendations
This paper creates an overall TFP growth decomposition model to estimate TFP growth at different economic levels, and decomposes overall TFP growth into technical effect, structure effect and the structural effect of capital and labor. Our findings are as follows:
(1) Since reform and opening up in 1978, China has maintained a high quality of economic growth thanks to its late- mover advantage, and about 1/ 3 of growth momentum stemmed from a general improvement in the level of technology in various sectors. In the process of China’s industrialization and urbanization, capital and labor concentrated in secondary and tertiary industries. The structural effect has
also supported China’s economic growth to some extent.
(2) After 2000, especially 2005, potentials of China’s late-mover advantage diminished, and the quality of its economic growth declined, as manifested in the falling contribution of technology progress to economic growth. On the other hand, structural effect and especially the structural effect of capital contributed an increasing share to TFP growth. This implies that against the backdrop of narrowed technology gaps, an effective way for China to support TFP and macroeconomic growth is to promote supply-side structural reforms and guide the flow of more factor resources from more productive sectors.
(3) Technical effect and structural effect played increasingly different roles in supporting industrial growth of various sectors. Growth of primary industry was mainly supported by technology progress. After 2000, technology progress independently supported the growth of primary industry. Technology progress in sub-sectors contributed about 1/3 of growth in secondary and tertiary industries, and the average contribution of structural effect was negligible. After 2005, however, the supportive role of technology progress swiftly diminished and even turned negative, and structural effect made a significantly positive contribution to industrial growth. After the eruption of global financial crisis in 2008, there was an obvious decline in the quality of China’s economic growth, which was primarily supported by inefficient inputs. Growth quality deterioration was particularly significant for secondary industry. This trend started to reverse around 2014.
( 4) For sub- sectors, those of secondary industry with excess capacity such as iron and steel, cement, electrolytic aluminum, flat panel glass and shipbuilding went through a significant technology stagnation or retrogression. In tertiary industry, finance and insurance and real estate experienced an even more serious long-term stagnation or retrogression in the level of technology. Factor allocation and concentration demonstrated an obvious tendency of reverse technology progress.
We have put forward the following policy recommendations:
(1) We should face the reality of China’s falling overall TFP growth and its falling contribution to economic growth in the post-crises era, draw lessons from the negative impact of stimulus policy and factor-driven growth, and balance the relationship between growth speed and growth quality. (2) In the short term, the structural effect of TFP growth should be brought into full play. In supply-side structural reforms, China should continue to phase out backward capacities in sectors with excess capacity, reduce bubbles in financial and real estate sectors, and offer reasonable incentives to guide the flow of factor resources to tech-intensive and efficient sectors. (3) In the mid- and long-term, technology progress would remain the primary source of TFP growth. China should implement the Outline of National Innovation-Driven Development Strategy, promote technology progress in various sectors, and foster technology advantages to undergird economic growth.