Total factor productivity in Chinese manufacturing firms: the role of E-commerce adoption

Abstract

This paper seeks to investigate the relationship between e-commerce and total factor productivity (TFP) at the manufacturing firm level. Using data of 178 A-share listed companies in China during the period from 2015 to 2021, the article empirically tests the questions of whether and how e-commerce used directly by manufacturing firms can boost their productivity growth. The result shows that the penetration of e-commerce in manufacturing has a positive and significant effect on TFP growth. It is also found that the spillovers of intra-firm human capital and the effect of inter-firm market competition both play a crucial role in linking e-commerce to firm-level TFP growth. Specifically, e-commerce is beneficial to driving TFP growth significantly through attracting high-quality rather than low-quality human capital accumulating in manufacturing firms, as well as through improving appropriately market concentration rather than increasing the intensity of market competition. This article contributes to the existing literature by exploring the productivity-driving force of manufacturing firms which face transformation from a focus on the production to the marketing chain within the context of online business. It also provides policy applications how to improve the TFP of manufacturing firms through the use of digital platforms.

Appendix: Estimating details concerning the OP and LP estimators

The Olley–Pakes (OP) method and Levinsohn–Petrin (LP) method, as two classical methods to measure the total factor productivity of enterprises, can overcome the potential endogeneity problem and control the loss of effective information quantity [58]. Therefore, this paper measures the TFP of 178 manufacturing A-share listed companies in China from 2015 to 2021 using the OP method and LP method. The OP or LP method is basically the same and the LP method is improved on the basis of the OP method. The measurement process is as follows.

First, the equation between the firm’s current capital stock and investment is constructed:

$$K_{it + 1} = (1 - \delta )K_{it} + I_{it}$$

(6)

where $K$ is the capital stock of the enterprise and $I$ represents the current investment.

Second, the basic equation, Ln Y = Ln A + α Ln K + β Ln L, is transformed into an econometric model:

$$\ln Y_{it} = \alpha \ln K_{it} + \beta \ln L_{it} + \omega_{it} + e_{it}$$

(7)

where ω_it is the part of the residual term that can be observed by the firm and affects the factor selection in the current period. e_it is the part of the residual term that contains unobservable technology shocks and measurement errors. After that, an optimal investment function is generated as follows:

$$i_{it} = i_{t} \left( {\varpi ,\ln K_{it} } \right)$$

(8)

We can find the inverse function of this optimal investment function. Assuming that h(·) = i⁻¹(·), ω can be written as

$$\varpi_{it} = h_{t} \left( {i_{it} ,\ln K_{it} } \right)$$

(9)

Substituting Eq. (9) into Eq. (7), we obtain the following equation:

$$\ln Y_{it} = \alpha \ln K_{it} + h_{t} \left( {i_{it} ,\ln K_{it} } \right) + \beta \ln L_{it} + e_{it}$$

(10)

It is defined that $\phi_{it} = \alpha \ln K_{it} + h_{t} \left( {i_{it} ,\ln K_{it} } \right)$, i.e.$\phi_{it}$ is a polynomial containing the logarithmic values of the investment and the capital stock, and define its estimate as $\mathop \phi \limits^{\sim }_{it}$, so that the following equation can be obtained as:

$$\ln Y_{it} = \phi_{t} + \beta \ln L_{it} + e_{it}$$

(11)

By estimating Eq. (11), we can obtain a consistent unbiased estimation coefficient of the labor term $\mathop \beta \limits^{ \wedge }$. Afterwards, the estimated coefficients are used to fit the value of the polynomial $\mathop \phi \limits^{\sim }_{it}$ consisting of the investment and the capital stock.

Finally, we define $V_{it} = \ln Y_{it} - \mathop \beta \limits^{ \wedge } \cdot \ln L_{it}$, and estimate the following equation

$$V_{it} = \alpha \cdot \ln K_{it} + g\left( {\phi_{t - 1} - \alpha \cdot \ln K_{it - 1} } \right) + \mu_{it} + e_{it}$$

(12)

where g(·) is a function containing ϕ and lags of the capital stock that can be estimated by higher-degree polynomials of ϕ_t−1 and Ln K _t−1. In the actual estimation process, it is necessary to use the nonlinear least squares method to ensure that the estimated coefficients of the capital stock are always consistent, both in the current period and in the lagged period. When the estimation results of Eq. (12) are obtained, all coefficients in the production function can be computed. This allows us to obtain the logarithm of the residuals by fitting the equation Ln Y = Ln A + α Ln K + β Ln L to obtain the logarithm of total factor productivity.

As stated above, the OP method can obtain the consistent estimates of total factor productivity at the firm level. However, the method assumes that the investment and total output need to maintain monotonical relationship at all times, so sample firms with zero investment cannot be estimated. In reality, some firms do not have positive investment every year, so using the OP method requires removing these firms from the sample. The improvement of the LP method is using the intermediate goods input to replace the investment as the proxy variable, which can avoid the problem that the TFP of the firm cannot be estimated due to zero investment.

According to the above methods, we obtain the data of A-share listed manufacturing enterprises in China to measure the TFP of firms. The paper uses operating incomes to represent total output, net fixed assets to represent capital input, and total number of employees to represent labor input. In addition, in the OP method measurement, the investment is represented by cash paid for the purchase and construction of fixed assets, intangible assets and other long-term assets; in the LP method measurement, the intermediate goods input is represented by “operating costs + selling expenses + administrative expenses + financial expenses − depreciation and amortization − cash paid to employees and cash paid for employees”.