Social Mobility and Firms’ Total Factor Productivity: Evidence from China

Abstract

The key to achieving sustainable growth is to increase total factor productivity. What role does the openness of social structure play in promoting total factor productivity? Based on data from the China Labor-force Dynamic Survey (CLDS), we used the methodology of intergenerational order correlation to measure city-level absolute mobility and to examine the impact of social mobility on the total factor productivity of Chinese manufacturing firms. The results show that the following: (1) Higher social mobility has a significant positive impact on firms’ TFP. Our results still hold after using the instrumental variables strategy to mitigate the endogeneity problem and performing a series of robustness tests. (2) Heterogeneity tests show that the positive effect on total factor productivity is more pronounced in non-state-owned enterprises, small- and medium-sized enterprises (SMEs), and technology-intensive enterprises. Meanwhile, a higher level of marketization reduces the sensitivity of enterprises’ TFP to social mobility. (3) Three identified channels are human capital allocation, fairness perception, and technological innovation. Our findings inspire the government to promote the system and mechanism reform of social flow and improve sustainable development ability.

1. Introduction

Since the 1960s, OECD countries represented by the United States have transferred low-end and mid-end industries to developing countries on a large scale through deindustrialization and production outsourcing [1]. Since then, emerging economies represented by China have gradually established the international competitiveness of manufacturing in the world through their endowment advantages. By 2019, the added value of China’s manufacturing industry accounted for 28.1% of the world’s total, maintaining a first place for 10 continuous years, which has greatly boosted economic growth. However, after the outbreak of the international financial crisis in 2008, developed economies have reawakened the importance of the manufacturing industry in long-term economic growth and their reindustrialization strategy may accelerate and weaken China’s competitiveness [2]. By adopting technology control or improving production efficiency through modern manufacturing technology, they strengthen their advantage across the entire manufacturing chain. As a result, China’s technological catch-up is being stymied and industrial development paths are in danger of being blocked. Faced with these challenges, China needs to achieve industrial upgrading and sustainable development through independent innovation. The success of innovation is ultimately measured by whether it can improve total factor productivity [3,4].

How to improve total factor productivity is a major concern of all economies. According to the data released by the World Bank from 2008 to 2016, the global average annual growth rate of total factor productivity was −0.4%, of which developed countries were close to zero. Thus, they have implemented a battery of economic policies, such as tax abatement and protectionist trade, aimed at revitalizing local manufacturing by accelerating technological innovation and enhancing the economy’s resilience to the crisis. In China, the average annual growth rate of manufacturing total factor productivity during the 13th Five-Year Plan period is about 2.2%, which generally remained at a low level. Nevertheless, total factor productivity still has great potential for improvement and is considered to provide strong support for China’s sustained economic growth [5,6,7]. A large amount of literature has been accumulated on the study of China’s total factor productivity but there is still no consensus. These disputes mainly focus on two aspects, the first of which is how to explain China’s economic growth. The author of [8] believes that it depends on government-led fixed asset investment but the low TFP growth rate has caused many scholars to worry about the sustainability of economic growth [9,10]. In contrast, the author of [11] stated that China’s GDP growth can be explained by the increase in total factor productivity, which contributes more than 60%. Especially after joining the WTO, the total factor productivity of Chinese industrial enterprises has greatly improved [12,13,14]. The second aspect is the source of China’s TFP growth. Some studies suggest that China’s TFP improvement obtains benefits from capital reallocation brought about by financial reform and state-owned enterprise reform but this view does not seem to be unanimously agreed upon by other scholars [15,16]. On the contrary, much of the literature tends to interpret it as technological progress caused by stimulating enterprise vitality [17,18,19]. An open social structure can fully mobilize the initiative of micro-individuals, resulting in better economic outcomes [20].

Social mobility refers to the relationship between a child’s achievement and family background, which is usually used to measure the openness of social structures [21,22]. In general, low social mobility means that it is hard for people to compete fairly and birth background is important. The resulting panic about the future ultimately reduces economic vitality. If social mobility is sufficient to increase the probability of upward mobility for elites in disadvantaged groups so that they can escape the intergenerational transmission of poverty by their own efforts rather than parental achievements, it can motivate the work effort and innovation enthusiasm of social members. As we know, production efficiency is not only inherent in the operation of factories or enterprises but it is also closely related to external factors beyond one’s control [23]. These external factors also include social mobility. Although social mobility may not directly operate on productivity levels, it can influence producers’ efforts and decision-making motivations. We also note that its incentive effect is related to whether producers can successfully move to a higher position within the industry’s productivity distribution. At present, the economic outcomes of social mobility, which are closely linked to the social structure, have been confirmed by many studies [24,25,26] and the relationship with enterprise productivity is still not well-known.

To fill the gap, we measured social mobility and investigates its impact on total factor productivity, following the work in [27]. As the author of [28] found, in Sweden, which has a high level of social mobility on the whole, there are obvious differences in social mobility among regions within the country. To improve the accuracy of our findings, we therefore focused on social mobility at the city level. Possible contributions to the existing literature are listed below. First, we extended the existing research perspective. Previous studies have mainly focused on measuring relevant indicators and discussing the dynamics of social mobility [29,30]. We quantitatively analyzed the impact of social mobility on firm-level TFP based on micro-data, providing empirical evidence for the government to issue policies that promote the social mobility of labor force and talent. Second, we analyzed how social mobility affects firm-level TFP from three channels, which not only deepen our recognition of social mobility but also help us to understand the relationship between the macro-environment and corporate activities from a broader perspective. Finally, in this paper, we confirm the positive effect of higher social mobility on improving total factor productivity and provide theoretical guidance for emerging economies to cope with the challenges caused by the reindustrialization of OECD countries. With the gradual disappearance of China’s demographic dividend, economic growth needs to shift from relying on capital and labor input to relying on increasing total factor productivity. By optimizing the incentive mechanism, enhancing the vitality of the microcosmic body is helpful to improve firms’ total factor productivity. From the perspective of policy implications, our research highlights the importance of social mobility in enhancing total factor productivity and promoting sustainable economic development.

The remainder of this paper is organized as follows: Section 2 introduces the theoretical analysis and research hypotheses. Section 3 describes the data and empirical methodology used in the study. Section 4 presents the benchmark results and our examination of the heterogeneous effects of social mobility on total factor productivity. Section 5 concerns the empirical analysis of the possible influencing mechanism and Section 6 concludes the paper.

2. Theoretical Analysis and Research Hypothesis

The position of productivity distribution within the industry determines whether a company has sufficient competitiveness to survive in the fierce market competition. What accounts for the widespread productivity differences between firms? A large amount of literature has discussed this issue and the research conclusions can be summarized as follows: (1) From the perspective of enterprises outside, various shocks of the external environment, such as trade frictions, exchange rate fluctuations, environmental regulations, and industrial policies, increase the uncertainty of production activities [31,32,33,34]. When the property rights of essential productive factors are transferred from factors’ owners to firms, the intensified uncertainty increases transaction costs and hinders the optimization of factor usage, thus reducing firm productivity. (2) From the perspective of enterprises inside, managerial capacity, staff quality, and technological innovation are critical to upgrading enterprises’ total factor productivity [23,35]. With a given salary level, whether a company can recruit high-skilled managers and employees for operation and technology R&D not only depends on outstanding talent storage in the labor market but also on the individual’s career decision in a specific social environment.

It is generally accepted that the social environment profoundly affects individuals’ behavior and decision making. On the one hand, as it pertains to this paper, a fair social environment frees enterprises from the dilemma of occupying limited funds to engage in rent seeking and alleviates the interference of non-market forces on corporate activities. On the other hand, regions with higher social mobility provide more opportunities for people to fairly compete, sending an incentive signal that efforts can be reasonably rewarded. By virtue of their ability and effort, people have the opportunity to obtain higher income and social status and thus achieve upward mobility. A rational workforce with the potential for upward mobility thus chooses to flow in. In this context, enterprises only need to concentrate on production and operating activities so that they can recruit many excellent managers and employees to improve production efficiency. Therefore, we propose the following hypothesis:

Hypothesis 1 (H1).

Higher social mobility contributes to increasing the total factor productivity of local firms.

There is a universal consensus among economists that economic growth cannot be separated from the public services provided by the public sector and productive activities undertaken by the private sector. Both sectors thus need substantial talents to optimize human capital [36,37]. As we know, the labor force is not only the main body of economic activities but it is also the accessory carrier of human capital. Optimizing the allocation of human capital indicates that high-tech physical capital should be matched with a high-level labor force, which contributes to improving labor productivity and promoting economic development. Nevertheless, this kind of situation seems to rarely happen because, given the total amount of human capital in the whole society, the occupation of human capital in any sector may crowd out another sector. So how is human capital allocated between the two departments? This is connected with individuals’ vocational choices. In fact, choosing an occupation hinges on the return ratio of capacity and whether it is linked to individuals’ efforts. People in environments with higher social mobility are more likely to obtain income and social status through their own abilities rather than family background. Assuming that the interference of the economic cycle is not taken into account, it is difficult for the public sector with a relatively fixed rate of return to attract high-quality talents who are confident in their efforts and value. The influx of advanced labor forces optimizes the human capital structure of the private sector, which is conducive to improving enterprise TFP. In contrast, lower social mobility may reduce, to some extent, the rewards people deserve in the private sector, triggering a rush of talented persons into the public sector, who are not directly involved in productive activities. Human capital mismatch arises when the talent supply in the private sector is severely squeezed, as confirmed by the author of [38]. The authors of [39] also found that the deterioration of human capital allocation was detrimental to enterprise efficiency. Based on the above analysis, the second hypothesis is as follows:

Hypothesis 2 (H2).

Social mobility may affect firms’ total factor productivity by adjusting the allocation of human capital between the government and enterprises.

Many studies have shown that social mobility has a significant impact on the public’s sense of fairness [40,41]. The initiation of justice implied by high social mobility makes people deeply realize that they can succeed through hard work. When competent people do not achieve a reasonable return on their efforts, a sense of injustice arises. The author of [42] also found that in countries with increased social mobility, people were more convinced that inequality was caused by justice. From an enterprises’ perspective, assuming that justice is the basic constraint condition for making decisions, managers and employees may reduce their work input in the production and operating activities when perceiving unfairness. Especially for manufacturing workers, the negative work attitude reduces production efficiency, even if enterprises invest enough resources. Therefore, we propose the following hypothesis:

Hypothesis 3 (H3).

Social mobility may affect firm-level total factor productivity by changing employees’ fairness perception.

Technological innovation is the main driving force of corporate success and contributes to improving production efficiency [43]. Enterprises in the fierce market competition can rely on innovation to obtain excess profits. High investment and high risk also force enterprises to be cautious in the decision making portion of technology R&D. In addition, the deteriorating social environment, such as unfairness, aggravates the difficulties faced by disadvantaged enterprises in financing and cost management, thus magnifying innovation risks. It should also be noted that being a civil servant in the institutional background of China seems to have a higher social assessment than being a corporate innovator. Lower social mobility signifies that the difficulty of obtaining innovation revenue widens the occupational class gap and drives a great quantity of talents into the public sector as a profit divider [44]. As a result, the innovative talent shortage is bound to lead to lower innovation output, ultimately delaying the improvement of production efficiency. Thus, we posited:

Hypothesis 4 (H4).

Social mobility may also affect firms’ total factor productivity through the channel of technological innovation.

3. Research Design

3.1. Data Source

In this study, A-share listed manufacturing companies from 2008 to 2019 were selected as research samples and screened as follows: First, we removed the samples of companies marked as ST and PT. Second, observations with inconsistent registered addresses and office addresses were excluded. Third, samples were also removed if data for a variable are missing. After completing the sample screening, we matched companies with social mobility at the city level based on registered addresses. The data used in this study are drawn from the following three sources: (1) The data of the China Labor-force Dynamic Survey (CLDS) are the first to adopt the tracking method of rotating samples, which better adapts to the Chinese environment with drastic changes and provides reliable data for this study. Using pooled cross-sectional data from 2012 and 2014, we attempted to measure social mobility at the city level. (2) Firm-level data are collected from CSMAR and Wind databases. (3) City-level data are obtained from the China City Statistical Yearbook.

3.2. Variables Description

3.2.1. Total Factor Productivity

As for long-term business development and sustainable economic growth, total factor productivity is important. This study exploits OP and ACF methodologies to calculate total factor productivity. The reasons are shown, as follows: (1) The authors of [45] proposed that enterprise investment was used to represent the proxy variable of unobtainable productivity, which overcomes the endogeneity between factor inputs and production efficiency. (2) In China, many companies enter or exit the A-share market every year, so the collected firm-level data are unbalanced. Compared to LP methodology, OP methodology is believed to eliminate sample selection bias caused by unbalanced panel data and enterprise withdrawal, although it is prone to sample loss. (3) As described by the authors of [46], ACF methodology could alleviate the collinearity of OP methodology by introducing labor input into the intermediate input function. Moreover, China’s relatively backward regulations concerning labor protection, to some extent, relax the strict assumption of high labor adjustment costs in enterprises, suggesting the feasibility of ACF methodology.

It is should be stated that when we calculate TFP, the firm’s operating income, net fixed assets, and cash payments to employees denote output (Y), capital (K), and labor force (L), respectively. Meanwhile, cash used to purchase fixed assets and intangible assets is considered a proxy variable for investment (I). The statistical characteristics of firms’ total factor productivity are shown in

Table 1. Summary statistics.

Variable	Mean	Std	Min	Max
Panel A: Firm level
TFP_OP	3.5014	0.6192	0.1086	6.8152
TFP_ACF	2.2772	0.5918	−1.0505	6.0389
ABM	2.5230	0.340	1.9576	3.1434
ROE	0.0635	0.2584	−10.3614	3.6214
ER	−0.5723	0.4349	−7.3509	1.0288
Size	7.6303	1.2351	2.0794	12.4380
SOE	0.3096	0.4624	0	1
Age	8.3977	6.5030	0	29
Panel B: City level
GRP	4.5858	0.0264	4.3138	4.6219
FEX	12.5668	1.9195	3.7842	15.5293
PPS	5.9657	0.8666	2.0794	7.8156

.

3.2.2. Social Mobility

Accurately measuring social mobility at the city level is a vital issue in this paper. Most of the previous literature took intergenerational income elasticity as a proxy variable but it was questioned by many scholars [25,47]. The main reason for this is that calculating intergenerational income elasticity requires tracking personal income data over time. Given the unavailability of high-quality data, life cycle bias is inevitable if we rush to replace permanent income with current income. To break through data limitations, the authors of [48] proposed measuring social mobility by employing directional status change. In contrast, the intergenerational order correlation is more suitable for comparison between regions.

By definition, social mobility can be divided into relative mobility and absolute mobility. Relative mobility refers to the relative results of children from high-income and low-income families based on their parents’ background. The specific groups are upward mobility and downward mobility. Absolute mobility emphasizes the expected rank of social status a child can achieve with similar parental achievement. If increased relative mobility comes at the cost of worse outcomes for children from low-income families, it is not desirable. For this reason, absolute mobility as a proxy variable is quite reasonable. In this study, based on the fact that social status is more comprehensive and reliable [49], we used subjective social status instead of income level to measure city-level social mobility. Equation (1) can be expressed as:

$$R_{ic}={\alpha }_c+{\beta }_c{P}_{ic}+{\epsilon }_{ic}$$

(1)

where $R_{ic}$ and $P_{ic}$ represent the social status of children and parents, respectively, in city c. Specifically, we also used their family social status at the age of 14 as a parental status ranking in the questionnaire, as well as extracting information about respondents’ current status ranking. ${\beta }_c$ represents the slope of the rank–rank relationship, measuring the degree of relative mobility in city c.

After separating ${\alpha }_c$ and ${\beta }_c$ from Equation (1), we then exploited Equation (2) to calculate absolute mobility at the city level. Equation (2) is given by:

$$r_{pc}={\alpha }_c+{\beta }_{c}p$$

(2)

where $r_{pc}$ is defined as the expected rank of a child who lives in city c with parents whose social status ranks in the p percentile, called absolute mobility. It is worth emphasizing that a sustainable development society should have necessary opportunities for fair competition and selection mechanisms so that the middle-low class, especially elites from vulnerable groups, have the opportunity to upward flow. If the possibility of obtaining the corresponding political and economic treatment through hard work is forcibly taken away, social vigor certainly is weakened, hindering sustainable development. Higher social mobility also contributes to raising allocation efficiency and dynamic equality. Consequently, the desire of the bottom group to change their status is taken into account when the rulers make policies. In this study, we focus on the average absolute mobility of the middle-low social classes (25th percentile), named ABM. Furthermore, the p percentile is replaced by the 10th percentile in subsequent robustness checks.

3.2.3. Control Variables

To control the other factors that influence the total factor productivity of enterprises as much as possible, referring to [50], we selected a series of control variables at the firm level and city level. Firm-level variables include the following: (1) Return of equity (ROE) is used to measure enterprise growth capability. (2) Shareholders’ equity ratio (ER) represents the asset structure. (3) Firm size (Size) is measured by the number of employees. (4) SOE denotes corporate ownership property. (5) Age is defined as firm age. In addition, city-level variables include the following: (1) We define GRP as the proportion of secondary and tertiary industries, indicating the regional industrial structure. (2) FEX is expressed as the logarithm of annual scientific research and technology expenditure at the city level. (3) PPS is measured by city-level population size. Likewise, we mitigated the interference of extreme values on the estimated results by winsorizing the continuous variables at 1%. Table 1 presents the descriptive statistics of all the variables.

3.3. Empirical Model

To investigate the impact of social mobility on firms’ total factor productivity, we exploit the following model as shown in Equation (3).

$$TF{P}_{i,j,c,t}={\alpha }_0+{\beta }_{1}AB{M}_{c,p}+{\displaystyle \sum {\beta }_m{X}_{i,j,c,t,m}+{\sigma }_t+{\sigma }_j+{\epsilon }_{i,j,c,t}}$$

(3)

where i is the enterprise, j is the industry, c is the city, and t is the year. $TF{P}_{i,j,c,t}$ measures total factor productivity at the firm level. $AB{M}_{c,p}$ is the independent variable, denoting that the absolute mobility of city c at the p percentile. $X_{i,j,c,t,m}$ represents several control variables. ${\sigma }_t$ and ${\sigma }_j$ are the year fixed effect and industry fixed effect, respectively. ${\epsilon }_{i,j,c,t}$ is the error term.

4. Empirical Results

4.1. Baseline Results

Table 2. Results of baseline regression.

Dependent Variable	TFP_OP		TFP_ACF
	(1)	(2)	(3)	(4)
ABM	0.0668 *** (0.0140)	0.0864 *** (0.0157)	0.0934 *** (0.0145)	0.0880 *** (0.0160)
ROE		0.9476 *** (0.0830)		0.9179 *** (0.0813)
ER		−0.3226 *** (0.0239)		−0.3151 *** (0.0234)
Size		0.0850 *** (0.0071)		0.0235 *** (0.0070)
SOE		−0.0982 *** (0.0141)		−0.1179 *** (0.0140)
Age		0.0086 *** (0.0011)		0.0073 *** (0.0011)
GRP		0.6420 *** (0.2411)		0.5268 ** (0.2379)
FEX		0.0047 (0.0055)		0.0015 (0.0055)
PPS		0.0188 ** (0.0092)		0.0113 (0.0092)
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Obs	11841	11272	11841	11272
R2	0.1804	0.3321	0.1907	0.2762

Notes: Standard errors in parentheses; ***, ** indicate the significance of 0.01, 0.05 respectively.

presents the estimated results based on the empirical model introduced in Section 3. In Column 1 and Column 2, the dependent variables are expressed as firm-level total factor productivity calculated by OP methodology. To be specific, Column 1 shows the impact of social mobility on firms’ total factor productivity when only year fixed effect and city fixed effect are controlled. The coefficient of social mobility is 0.0668 and is significant at the 1% level, indicating that higher social mobility contributes to improving the total factor productivity of local enterprises. Column 2 reports the results after adding firm-level and city-level controls to the regression model. It can be found that one standard deviation increase in social mobility causes an 8.64% increase in firms’ TFP, which is consistent with H1. The estimated results of control variables are basically in line with our expectations. Among them, return of equity, firm size, and firm age all have a positive impact on enterprise total factor productivity. Except for scientific research and technology expenditure, other variables in city-level controls are beneficial to improving firms’ TFP.

In Columns 3–4, we took firm-level TFP measured by the ACF methodology as the dependent variable to conduct robustness tests. As shown in Column 3, the estimated coefficients of absolute mobility and total factor productivity are significantly positive before adding control variables. Moreover, we added firm-level and city-level variables, simultaneously controlling for the year and city fixed effects in Column 4. It can be seen that the coefficient of social mobility is 0.088 and is significant at the 1% level, implying that the baseline results are robust. In conclusion, our findings provide empirical evidence for the influence of social environment changes on corporate productivity.

4.2. Robustness Checks

4.2.1. Endogenous Concerns

Ignoring endogeneity issues may lead to a bias in estimating results. Therefore, we conducted robustness tests from the following perspectives.

First, consider the influence of unobserved variables on the estimating results. Although we control for a battery of firm-level and city-level variables, there may be unobserved variables, such as culture, values, and religion, that lead to erroneous conclusions. In recent years, more and more economists have paid attention to the economic consequences of culture. While micro-individuals (including corporate managers and employees) live in a specific cultural environment, their behaviors and decisions are subtly influenced. Numerous studies have provided evidence to support this viewpoint [51,52]. Given the fact that China has distinct regional differences, cultural diversification is likely to affect the activities of local enterprises. Hence, according to the measurement method of culture proposed by the authors of [53], this paper selects cultural indicators that may be theoretically related to independent variables or dependent variables to characterize regional culture. Specifically, power gap (PD) is defined as the extent to which social members accept that power can be shared unequally among all of us, and mastery (Mastery) describes the extent to which individuals and groups should change their social environment. Furthermore, referring to [54], we also use the city-level number of Qing Dynasty academies (AA) taken from the Chinese Research Data Services (CNRDS) platform to control the influence of social values. The results in

Table 3. Results of controlling culture and value.

Dependent Variable	TFP_OP			TFP_ACF
	(1)	(2)	(3)	(4)	(5)	(6)
ABM	0.1063 *** (0.0169)	0.0722 *** (0.0162)	0.0875 *** (0.0155)	0.1062 *** (0.0170)	0.0771 *** (0.0166)	0.0887 *** (0.0159)
PD	0.1262 ** (0.0295)			0.1131 *** (0.0295)
Mastery		−0.1440 *** (0.0523)			−0.1095 ** (0.0520)
AA			0.0242 *** (0.0032)			0.0216 *** (0.0032)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Obs	11,272	11,272	11,272	11,272	11,272	11,272
R2	0.2793	0.3326	0.3356	0.2212	0.2765	0.2794

Notes: Cultural indicators (PD and Mastery) at the province level are derived from [55], as detailed in Appendix A; standard errors in parentheses; ***, ** indicate the significance of 0.01, 0.05, respectively.

show that the coefficient of social mobility is still significantly positive after controlling the cultural effect, which confirms our findings.

Secondly, consider the measurement error of social mobility. Following common practice in the existing literature, we calculated absolute mobility at the city level by only retaining people who have not migrated. However, since the 1990s, there has been a large-scale population migration between Chinese cities, such as Beijing, Shanghai, Shenzhen, and other first-tier cities that have a high proportion of the migrant population. If absolute mobility is only measured with local population samples, the measurement errors in these cities may lead to an underestimation of our empirical results. To mitigate the possible interference of measurement errors in social mobility, we excluded the top 10 cities with net population flow and the top 5 provinces with population migration rates. As can be seen from the results in

Table 4. Results of controlling population migration.

Dependent Variable	Removing the Top 10 Cities with Net Population Flow		Remove the Top 5 Provinces with Population Migration Rate
	TFP_OP	TFP_ACF	TFP_OP	TFP_ACF
ABM	0.0877 *** (0.0171)	0.0891 *** (0.0175)	0.0984 *** (0.0305)	0.0932 *** (0.0305)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Obs	6440	6440	6292	6292
R2	0.3505	0.3030	0.3156	0.2668

Notes: Standard errors in parentheses; *** indicate the significance of 0.01.

, higher social mobility improves local firms’ total factor productivity, suggesting that the benchmark results are robust.

Thirdly, consider controlling for private economic development. Since the market reform in 1993, private enterprises have played an irreplaceable role in promoting regional economic growth in China. In particular, regarding price reform, market competition, and inter-regional flows of factors, private enterprises are important. Many studies have shown that vibrant private economies profoundly influence local enterprise culture, business climates, and job market structures, which have the potential to change the regional social structure [56,57]. Thus, we exploited non-state-owned economic development scoring at a provincial level in the marketization index from 2008 to 2019 as a proxy variable, including the proportion of the non-state-owned economy in prime revenue (Psoe1), total investment in fixed assets (Psoe2), and employment (Psoe3) of industrial enterprises. As shown in

Table 5. Results of controlling for differences in the private sector.

Dependent Variable	TFP_OP			TFP_ACF
	(1)	(2)	(3)	(4)	(5)	(6)
ABM	0.0790 *** (0.0157)	0.0940 *** (0.0158)	0.0791 *** (0.0155)	0.0807 *** (0.0160)	0.0954 *** (0.0161)	0.0815 *** (0.0159)
Psoe1	0.0524 *** (0.0142)			0.0507 *** (0.0140)
Psoe2		0.2664 *** (0.0289)			0.2640 *** (0.0288)
Psoe3			0.2132 *** (0.0287)			0.1850 *** (0.0285)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Obs	11,272	11,272	11,272	11,272	11,272	11,272
R2	0.3329	0.3372	0.3351	0.2771	0.2817	0.2787

Notes: Standard errors in parentheses; *** indicate the significance of 0.01.

, the coefficients of social mobility are significantly positive at the 1% level, which is consistent with the baseline regression results.

Finally, consider instrumental variable strategy. Regarding corporate activity, although the regional social structure is relatively exogenous, there may still be a reverse causality. The authors of [58] found that the rapid growth of enterprises can not only improve local economic benefits but also change class mobility. Due to the relatively low status of merchants in traditional Chinese society, people took a special approach to pursue wealth. In other words, they first gained social status through imperial examinations and then seized wealth by virtue of their status. Especially for the common class, their disadvantages in status and capital forced them to succeed in the imperial examination to achieve the class transition. In this context, the imperial examination system gradually formed a kind of social value through intergenerational transmission. Therefore, local social mobility is correlated with the imperial examination system. Referring to [59], we chose the city-level number of Jinshi in the Qing Dynasty as an instrumental variable to represent social mobility. In

Table 6. IV estimation.

Dependent Variable	First Stage	Second Stage
	ABM	TFP_OP	TFP_ACF
	(1)	(2)	(3)
ABM		1.0076 *** (0.1267)	0.9613 *** (0.1247)
Keju	0.1053 *** (0.0075)
controls	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes
Obs	11,272	11,272	11,272
R2	0.1411	0.1026	0.0513
F-values	61.17

Notes: Standard errors in parentheses; *** indicate the significance of 0.01.

, Column 1 displays the first-stage regression result based on the two-stage least square (2SLS) strategy. Keju’s estimated coefficient is significantly positive with F-values larger than 10, indicating that the regions affected by the imperial examination system have higher absolute mobility, which is consistent with the findings in [54]. According to the second-stage results in Columns 2–3, social mobility has a positive impact on local firms’ total factor productivity. Our estimates are still robust.

4.2.2. Additional Robustness Checks

Resetting the Research Sample

We measured social mobility at the city level by using the data of the China Labor-force Dynamic Survey (CLDS) in 2012 and 2014. Although the authors of [27] believed that city-level social mobility would not change dramatically in a short period, it still seemed difficult to explain enterprise productivity that occurred before 2012. For this reason, we only retained company samples after 2012 and the regression results are shown in Columns 1–2 of

Table 7. Additional robustness checks.

Dependent Variable	Remove the Sample of Companies before 2013		Full Sample
	TFP_OP	TFP_ACF	TFP_OP	TFP_ACF	TFP_OP	TFP_ACF
	(1)	(2)	(3)	(4)	(5)	(6)
ABM	0.0696 *** (0.0195)	0.0704 *** (0.0199)
ABM1			0.1489 *** (0.0372)	0.1500 *** (0.0373)
REM					0.0190 (0.0162)	0.0134 (0.0161)
controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Obs	7054	7054	11,272	11,272	11,272	11,272
R2	0.3390	0.2650	0.3312	0.2752	0.3302	0.2740

Notes: Standard errors in parentheses; *** indicate the significance of 0.01.

. The coefficients of social mobility remained positive and statistically significant, indicating that the baseline results are not affected by resetting the sample.

Replacing Independent Variable

Regarding social mobility, we focused on the middle-low class upward flow. From a societal perspective, the incentive mechanism is not distorted, which is only beneficial to long-term economic growth if a reasonable level of social mobility can dramatically motivate individuals’ enthusiasm at the bottom of the status distribution. Similarly, this study uses the city-level social mobility, measured by the average absolute mobility 10th percentile as an explanatory variable (ABM1). The results of Columns 3–4 in Table 7 show that social mobility contributes to upgrading local firms’ TFP, confirming our findings.

Furthermore, we also examined the relationship between relative mobility and firm total factor productivity. By definition, increased relative mobility may be caused by children’s upward flow from families with below-median income parents or result from the downward mobility of children growing up in high-income households. Since it is difficult to clearly identify, we infer that relative mobility (REM) is not correlated with firm productivity. In Columns 5–6, the estimated coefficients are positive but not significant, which once again verifies the rationality of choosing absolute mobility to represent the openness of social structure.

4.3. Heterogeneity Tests

The above analysis shows that higher social mobility improves local firms’ total factor productivity. Next, we investigated the heterogeneity relationship from four aspects, including state-owned enterprises and non-state-owned enterprises, large-scale enterprises and small-scale enterprises, technology-intensive enterprises and traditional industrial enterprises, and enterprises in regions with a higher marketization level and enterprises in regions with a lower marketization level.

4.3.1. Ownership Heterogeneity

Compared to private enterprises (PEs), state-owned enterprises (SOEs) have closer ties with local governments and easier access to tax incentives and credit resources. Meanwhile, in China, since state-owned enterprises’ behaviors and decisions are mostly in response to policy orientations and help the government undertake part of the social security function, they face less pressure from efficient competition. We should also note that changes in the external environment, such as social mobility, have relatively limited impacts as SOEs can provide employees with stable jobs and generous benefits. Hence, we constructed a dummy variable of firm ownership (private enterprises, SOE equal to 1) and introduced the interaction term SOE*ABM into Equation (3). Column 1 and Column 5 in

Table 8. Heterogeneity tests.

Dependent Variable	Firms’ Total Factor Productivity
	(1)	(2)	(3)	(4)
Panel A: TFP_OP
ABM × Soe	−0.1798 *** (0.0467)
ABM × Size		0.1387 *** (0.0049)
ABM × idus			−0.1115 * (0.0605)
ABM × Market				0.0255 *** (0.0052)
ABM	0.1096 *** (0.0170)	0.0167 (0.0146)	0.0928 *** (0.0152)	0.0752 *** (0.0150)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Obs	11,272	11,272	11,272	11,272
R2	0.3332	0.4735	0.3323	0.3335
Panel B: TFP_ACF
	(5)	(6)	(7)	(8)
ABM × Soe	−0.1837 *** (0.0467)
ABM × Size		0.1227 *** (0.0050)
ABM × idus			−0.1277 ** (0.0604)
ABM × Market				0.0193 *** (0.0052)
ABM	0.1115 *** (0.0173)	0.0262 * (0.0146)	0.0951 *** (0.0152)	0.0794 *** (0.0150)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Obs	11,272	11,272	11,272	11,272
R2	0.2775	0.3137	0.2734	0.2771

Notes: The dependent variables in Panel A and Panel B are based on the OP and ACF methods, respectively; standard errors in parentheses; ***, **, * indicate the significance of 0.01, 0.05, and 0.1, respectively.

report the results. The coefficients of the interaction term are significantly negative at the 1% level, indicating that private enterprises’ total factor activity is more affected by local social mobility.

4.3.2. Firm Size Heterogeneity

The resource-based theory holds that larger enterprises are effective at resource orchestration while smaller companies can only rely on resource patchwork. Among them, resource orchestration emphasizes that enterprises have the capacity to manage orderly internal and external resources on the basis of existing ones. In general, large enterprises have a strong ability to orchestrate resources. Even if the external environment they faced tends to be unfavorable, they can still effectively integrate and utilize orchestrated resources to strengthen corporate abilities. Correspondingly, due to resource endowment constraints, it is difficult for small or medium enterprises (SMEs) to effectively orchestrate existing resources. In this case, disadvantaged business enterprises can only passively cobble together existing resources to cope with changes in the external environment. To better understand the heterogeneous impact of social mobility on firms’ TFP with different sizes, we constructed a dummy variable of firm size according to the enterprise classification rules formulated by the National Bureau of Statistics. Table 8 reports the estimated results. As shown in Column 2 and Column 4, the interaction coefficients are positively significant, implying that social mobility has a greater impact on SMEs’ total factor productivity. Our findings are consistent with the view proposed by the authors of [60] that SMEs’ activities are more vulnerable to external environmental shocks.

4.3.3. Industrial Heterogeneity

According to the industry classification of the China Securities Regulatory Commission, the companies selected as research samples are divided into traditional industrial enterprises and technology-intensive enterprises. See Appendix B for more details. We can see from Column 3 and Column 6 of Table 8 that the estimated coefficients of the interaction terms are negatively significant, at least at the 10% level. The results imply that a social structure with higher openness improves the social factor productivity of technology-intensive enterprises more than that of traditional industrial enterprises. This may be because the inherent nature of technology-intensive industries requires enterprises to have high-quality innovative talents and higher innovation ability. If it is hard for people to earn a reasonable reward for their efforts, social incentives are distorted. Once highly qualified employees are desperate to be rent-seekers rather than producers, the desire for technological innovation is severely curbed. In the end, companies cannot boost productivity even if they invest a lot of talent and funds.

4.3.4. Marketization Heterogeneity

There is a general consensus that market-oriented reform can reduce the cost of resource allocation and improve economic efficiency. Different market-oriented environments have created a differentiated performance of enterprises in acquiring and utilizing resources. To be specific, the activities of enterprises, which operate in regions with a high degree of marketization, are mostly carried out in accordance with market rules and are less interfered with by the government. At this point, a fairer competition environment can provide space for firm growth, encourage enterprises to increase investment to upgrade productivity, and then win the market competition. We use the data collected from China Marketization Index Reports in 2016 to investigate the heterogeneous impact of marketization levels on the relationship between social mobility and firm TFP. In Table 8, Column 4 and Column 8 report the estimated results. The coefficients of interaction terms are significantly positive, indicating that a higher marketization level can better play the role of social mobility in improving enterprise productivity. Despite remarkable achievements in China’s market-oriented reform, there are still obvious differences in the process of marketization between regions. Only by eliminating the backward regional protectionism in the market-oriented reform can we form a resource allocation mechanism that is conducive to fair competition.

5. Potential Mechanisms

In this section, we examine three possible channels through empirical analysis to provide a more comprehensive explanation of how social mobility affects firms’ total factor productivity.

5.1. Human Capital Allocation

In theory, human capital allocation among departments is the result of individual career choices under existing constraints. What kind of occupation an individual prefers is determined by the relative reward structure that is related to the game rules in society. Faced with lower social mobility, it is difficult for individuals to achieve the same reward as those from superior family backgrounds, even if they try their best to make more of an effort. Thus, a highly qualified workforce is more eager to enter the public sector and even hopes to increase the rate of return through rent-seeking activities. The private sector is also unable to boost productivity because it is difficult to attract these talents.

To test the above mechanism, referring to [61], this study introduces a proxy variable. The variable Hcapital is defined as the city-level ratio of education level to government and enterprises, which is used to measure the inter-departmental human capital structure. Specifically, we used micro-individual data from the China Household Finance Survey (CHFS) in 2015 to classify individuals whose work units are government agencies and public institutions (in the public sector), while collective enterprises, private enterprises, and individual businesses were correspondingly grouped into the private sector. Then, the proxy variable of human capital allocation was obtained by assigning value to a given education level, as detailed in Appendix C. In Column 1 of

Table 9. Mechanism analysis: human capital and fairness perception.

Dependent Variable	Hcapital	TFP_OP	TFP_ACF	Fairness	TFP_OP	TFP_ACF
	(1)	(2)	(3)	(4)	(5)	(6)
Hcapital		0.3434 *** (0.0985)	0.2801 *** (0.0981)
Fairness					0.2863 *** (0.0518)	0.3100 *** (0.0516)
ABM	−0.0260 *** (0.0025)	0.0940 *** (0.0160)	0.0941 *** (0.0162)	0.0268 *** (0.0048)	0.0813 *** (0.0154)	0.0823 *** (0.0157)
Firm-level controls	No	Yes	Yes	No	Yes	Yes
City-level controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	No	Yes	Yes	No	Yes	Yes
Industry FE	No	Yes	Yes	No	Yes	Yes
Obs	11,614	11,272	11,272	11,614	11,272	11,272
R2	0.4726	0.3328	0.2767	0.2458	0.3340	0.2787

Notes: Standard errors in parentheses; *** indicate the significance of 0.01.

, the city-level regression results show that social mobility significantly reduces the proportion of local high-quality talents entering government departments, implying that a fair social government provides enterprises with abundant talent supply. The coefficients in Columns 2–3 are positive and significant at the 1% level. Therefore, we can assert that in cities with a higher proportion of human capital allocation between the government and enterprises, the sufficient supply of a highly qualified workforce intensifies the competition among enterprises and helps them successfully move to a higher position within the industry’s productivity distribution. H2 is supported.

5.2. Fairness Perception

The rational economic participant assumption holds that people with the nature of pursuing fairness are not only interested in their own benefits but are also concerned with the differences between their benefits and those of others. In this work, higher social mobility inevitably brings individuals a higher sense of fairness. From an enterprise’s perspective, employees are highly sensitive to the compensation dispersion between themselves and their colleagues. If others gain the same benefits simply by virtue of their birth background rather than their own efforts, the perception of fairness can be weakened. An aversion to unfairness reduces employee motivation, which may count against firms’ production efficiency.

After extracting information from the China Household Finance Survey (CHFS) in 2015, based on the question “How do individuals evaluate social justice?”, we subsequently assigned values to the five options. For example, “very fair” was assigned a value of 5 and “very unfair” is equal to 1. In this way, we can calculate social equity in the average sense, which can be used to measure perceived equity at the city level. As can be seen from Column 4 of Table 9, people living in cities with higher social mobility have stronger perceptions of social equity. The results of Columns 5–6 show that perceived fairness significantly improves enterprises’ total factor productivity. Changing perceived equity is one of the potential mechanisms by which social mobility affects firm productivity. Hence, H3 is confirmed.

5.3. Technological Innovation

In Section 2, we discussed that the highly qualified laborer’s choice is affected by the subjective utility of occupational social status in addition to income. Under the influence of Confucianism, working in government or public institutions has always enjoyed a high social status in China. In other words, civil servants are more respected in this society than other professions, such as corporate innovators, resulting in occupational class differences. In particular, innovative groups living in an environment of low social mobility often find it difficult to obtain their due economic returns or social status through their own efforts, which magnifies the already obvious differences between professional classes. At this time, the highly capable labor force is more inclined to work in the public sector, so the total amount of innovative talents also decreases. For enterprises, the reduction in the number of talents engaged in innovative activities means lower innovation ability and it is more difficult to improve enterprises’ total factor productivity.

Using the patent application count data from A-share listed manufacturing companies, we constructed a proxy variable for innovation ability in order to verify the discussed mechanism. Taking into account the different definitions of patents by the State Intellectual Property Office of China, we calculated the weighted patent counts of the sample companies each year by assigning weights of 1/2, 1/3, and 1/6 to inventions, utility models, and designs, respectively. The estimated results are reported in

Table 10. Mechanism analysis: technological innovation.

Dependent Variable	Innovation	TFP_OP	TFP_ACF
	(1)	(2)	(3)
Innovation		0.0133 *** (0.0040)	0.0084 ** (0.0040)
ABM	0.7427 *** (0.1751)	0.0855 *** (0.0156)	0.0873 *** (0.0159)
Firm-level controls	Yes	Yes	Yes
City-level controls	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes
Obs	11,272	11,272	11,272
R2	0.5547	0.3327	0.2765

Notes: Standard errors in parentheses; ***, ** indicate the significance of 0.01, 0.05, respectively.

. In Column 1, the coefficient of ABM is positively significant, indicating that higher social mobility enhances enterprises’ innovation capability. Similarly, according to the results of Columns 2–3, we can infer that social mobility improves firm-level total factor productivity by enhancing innovation capacity. H4 is supported by our empirical evidence.

6. Conclusions and Policy Implications

Social infrastructure is crucial to sustained and healthy economic development. In this paper, we quantified absolute mobility at the city level using the data of the China Labor-force Dynamic Survey (CLDS) and empirically examined the impact of social mobility on enterprises’ total factor productivity by taking A-share listed manufacturing companies as research samples. We found that higher social mobility contributes to improving local firms’ total factor productivity. After a series of robustness checks, our results are still supported. Heterogeneity analysis showed that the positive effect of social mobility on firm-level TFP is more salient in private enterprises, small- and medium-sized enterprises, and technology-intensive industries in a given city. A higher degree of marketization helps to play an incentive role in social mobility. In addition, we also find that social mobility shapes firms’ total factor productivity mainly through human capital allocation, fairness perception, and technological innovation.

Our study has some significant policy implications for reforms to improve the social mobility of labor force and talent. First, improving social mobility may be an alternative to China’s industrial upgrading, because lower social mobility distorts the social incentive mechanism, resulting in the inability to release social innovative vitality, which to some extent inhibits entrepreneurship. The government needs to optimize the institutional structure that can effectively eliminate the inefficient sectors so as to actively create opportunities for social mobility and stimulate the potential of talents. Second, regarding developing countries seeking economic growth, we should be alert to the talent misallocation trap caused by rent seeking. By creating a fair and open social environment, more talented people can be encouraged and attracted to enter the productive sector so as to achieve a sustainable development capacity. The government should also combine the mismatch of human capital with China’s economic restructuring and formulate appropriately phased strategies for human capital accumulation to avoid the waste of educational resources. Finally, to cope with the severe international situation, we also need to optimize the business environment and actively support private enterprises, especially technology-intensive enterprises, to boost the development of the manufacturing industry.

There are some limitations in this study. First, when examining the impact of social mobility on firms’ TFP, we limited the research sample to Chinese listed companies. However, compared with listed companies, a large number of industrial enterprises above the designated size do not need strict supervision. People who work in listed companies merely account for a small part of the total labor force. The next step is to expand the research sample to industrial enterprises above the designated size in China to explore whether interesting findings different from those found in this paper can be obtained. In addition, this study was conducted in China. We believe that future research can examine the impact of TFP in other emerging countries at different stages of economic development.