Does Service Trade Liberalization Promote Service Productivity? Evidence from China

Abstract

This paper investigates the effects of service trade liberalization on service productivity. Based on a city-level database from 2006 to 2019, we estimate the labor productivity of the service industry and show the varying trends of productivity growth. Additionally, by exploiting the cross-city, cross-time variation at the time of China’s innovation pilot city policy of service trade liberalization, we employ a difference-in-difference strategy to estimate the effects. The empirical evidence suggests that liberalizing the service trade has a positive effect on service productivity. In addition, the evidence indicates that service trade liberalization could increase the productivity of the service industry in cities located in the eastern and central parts of China, as well as in those cities with a higher degree of marketization. The main influencing channels of service trade liberalization on productivity predominantly occur through the increase in technology spillover, market competition, and human capital. China’s evidence highlights the positive relationship between trade liberalization and productivity in the service industry and provides implications for realizing the sustainable development of services in developing countries; therefore, China and other developing countries are suggested to continuously formulate and deepen their service trade liberalization strategy.

1. Introduction

Service trade liberalization has become an increasingly important component of international trade and a key path to achieving efficient economic sustainable development [1]. Since China became a member of the WTO, China has been paying attention to the productivity and quality of the service sectors. Services in China account for over 50 percent of the GDP, and the service trade has been the second largest service trade economy globally since 2014 [2]. However, the competitiveness of the service trade in China has always been at a low level, and the restrictiveness of the service trade is estimated to be at a high level compared with developed and other emerging economies [3]. Thus, with the concept of “trade-led growth”, in 2015 the Chinese government formulated the new strategy of service trade liberalization, aiming at promoting the productivity of services [4].

Research shows that service trade liberalization accelerates productivity, and it can stimulate real consumption by decreasing transaction costs and altering the production structure of firms [5,6]. Service trade liberalization can also be beneficial for the stimulation of trade [7,8], enabling firms to leverage the advantages of learning through exports and hence, augmenting productivity [9,10]. Much of the literature on service trade liberalization has focused on its effect on the productivity of manufacturing firms. Few studies, particularly in developing countries, have focused on service productivity. Service trade liberalization is of great significance to developing countries, as it provides access to external production factors and advanced technology, encourages industrial competition, and enhances productivity [2]. Despite the potential benefits of service liberalization, it can also lead to a decrease in the productivity of service firms in developing countries when competition intensifies, as evidenced by Lai and Chen [11] and Faber et al. [12]. As the largest developing country, China’s service industry is facing a number of challenges, such as weak competitiveness, the vulnerability of private firms, and a lack of awareness of service consumption. The entry of foreign service providers and the influx of investment capital is likely to have a detrimental effect on China’s service sector. Thus, will the liberalization of services in China result in a higher productivity of the service industry?

In this paper, taking advantage of the policy of China’s service trade liberalization pilot cities, we seek to explore this issue. This policy aimed to motivate the pilot cities to raise the level of trade liberalization in the services sector, allowing us to identify both the temporal changes before and after, as well as the differences between pilot and non-pilot cities. To identify the impact, by exploiting the cross-city, cross-time variation at the time of China’s innovative pilot city policy of service trade liberalization, we use a difference-in-difference (DID) strategy in this paper.

Our paper finds that China’s service trade liberalization has a positive impact on the productivity of the service industry. Compared with the findings of Papaioannou from a cross-country perspective [13], this paper shows the conclusions at the city level. Multiple robustness tests have been conducted to validate this finding. To dispel doubts about the conditional assumptions as Lechner [14] has discussed, we have conducted a pre-existing parallel test and a placebo test. To further verify the robustness of our baseline results, we also measure major variables using the total factor productivity [15] and foreign capital access to substitute the explanatory variable and the explained variable, respectively. In addition, the evidence indicates that service trade liberalization in China can increase the productivity of service delivery in cities located in the eastern and central parts of China, as well as in those with a higher degree of marketization. There is no discernible impact of service trade liberalization on the productivity of cities located in the western area of China or in cities that have a low degree of marketization. In addition, this paper has examined three primary influencing channels. First, refer to Mattoo et al. [16], etc., service trade liberalization has technology spillover effects. The liberalization of services draws more capital to fund research and development (R&D), thus facilitating the innovation of service industries and leading to technological spillovers that can boost productivity. Second, service trade liberalization has competition effects [17]. Liberalization of the service trade encourages more service firms to enter pilot cities. By introducing external service suppliers, the competition in the local service market intensifies, prompting service firms to increase their productivity. Third, service trade liberalization also has human capital effects [18]. Policy measures that liberalize the service trade in pilot cities can be used to encourage more high-skilled labor to take part in service activities. The enhancement of human capital leads to an increase in service labor productivity.

The findings of this paper contribute to the literature in three ways. First, this paper enriches the existing research on the relationship between trade liberalization and productivity, with a specific focus on services. The connection between goods trade liberalization and manufacturing productivity, and the connection between service trade liberalization and manufacturing productivity are identified by numerous papers, but few studies have addressed the connection between service trade liberalization and service productivity. Second, this paper expands upon the recent study of the effects of service trade liberalization. By estimating the service productivity at the city level, this research proves the impact of service trade liberalization policy on the development efficiency of the service industry at a city-level perspective. Third, this paper adds to the existing measurements of service trade liberalization. By virtue of the exogenous service trade policy imposed by the Chinese central government, we are able to differentiate between pilot cities and non-pilot cities at the city level, which creates a quasi-natural experiment, enabling us to use the difference-in-difference approach for empirical research.

The remainder of this paper is structured in the following manner: Section 2 provides a brief literature review of the relevant topics. Section 3 investigates the city-level policy of the China service trade liberalization and the advancement of China’s service industries. In Section 4, we outline the method employed and the variables selected. Section 5 contains the results of our empirical research, such as the baseline results, robustness tests, and heterogenous discussions. Section 6 delves into the fundamentals of how service trade liberalization impacts service productivity. Section 7 outlines the conclusions, shows the policy implications, and points out the limitations and potentials of this paper.

2. Literature Review

There is increasing evidence that service trade liberalization could improve economic efficiency, with potentially large payoffs. On the macrolevel, social welfare gains have been increased by allowing greater access to foreign services [5,6], and internal market expansion and changes in the extent of the market have been brought about due to increased opportunities for service trade [19]. The impact of a reduction in the tariff equivalents of services barriers by 33 percent brings 2 more percent of GDP gains for developed countries, and a 7 percent increase in developing countries such as Tunisia [20]. On the microlevel, firms also benefit from service trade liberalization. Service trade liberalization can improve productivity in manufacturing firms, boost firm’s competitiveness, and promote the imports and exports of intermediates and goods [7,8,21,22]. Additionally, the positive effects of service imports have been found to improve firm-level employment in Germany [23]. Even though service trade liberalization is one way to improve economic efficiency, service trade liberalization in developing countries has experienced little progression, as developing countries generally have a conservative attitude toward service openness [24]. Whether service trade liberalization will bring extra negative effects to services in developing countries is a frequently debated issue. On the one hand, technology spillovers from service trade liberalization may cause local, under-developed suppliers to leave the market due to higher competition [11]. On the other hand, the development of one sector within the service industry may encourage reallocations of economic activities and absorb resources through the market integration effect, which may lower other sectors’ productivity levels [12]. Considering the uncertain effect of service trade liberalization on the service industry in developing countries, this paper shows how service trade liberalization in China, the largest developing country in the world, has increased productivity in the service industry. The results of this paper confirm the positive impact of service trade liberalization on economic development efficiency in developing countries.

Many papers have emerged in recent years and discussed the relationship between trade liberalization and productivity. Initially, the literature examined the connection between goods trade liberalization and manufacturing productivity, as numerous scholarly papers demonstrated that trade liberalization had a negative impact on productivity growth in the manufacturing sector [25,26]. Recent studies have revealed that service trade liberalization has a noteworthy impact on the productivity of manufacturing firms, as demonstrated by Arnold et al. [27,28] and Beverelli et al. [29]. However, only a few studies have discussed the connection between service trade liberalization and service productivity. Papaioannou [13] examined the effects of service liberalization on service total factor productivity (TFP) growth across European countries, drawing upon data from the service industry. Unlike his research, our article attempts to evaluate the consequences of service liberalization on service productivity from an internal outlook, since it furnishes information at the city level.

Several papers have demonstrated that the liberalization of a single service sector can result in increased productivity. Lower-income countries do not have the institutional or regulatory structures to ensure the effective functioning of the financial sector. Eschenbach and Francois [30] argue that openness in the finance sector has a positive impact on growth. Lestage et al. [31] show that a higher quality and greater diversity of telecommunications services are associated with policies encouraging greater FDI and competition in the telecommunications sector. Fink et al. [32] find that trade liberalization in terms of transport will reduce transport prices. This research examines the overall outcome of the liberalization of service sectors and demonstrates its positive effects on service productivity at the city level.

To explore the impact of China’s service trade liberalization, it is necessary to measure the service trade liberalization and service productivity. The measure of service trade liberalization is different from goods trade liberalization. Unlike the barriers to the goods trade that can be quantified by tariffs, restrictiveness and barriers in the service trade depend on the necessary certifications of service providers and operational prohibitions, as noted by Kox and Lejour [33]. To measure service trade liberalization, de-regulation indices, such as the Service Trade Restrictiveness Index (STRI) [34,35], Foreign Direct Investment (FDI), Restrictiveness Index [36,37], and Product Market Index (PMR) [38], can be used. The impacts of liberalization have been investigated through service reform [12,28] and multilateral or regional trade agreements [6,39]. Our paper also employed the service trade policy to evaluate the degree of service trade liberalization. By virtue of the exogenous service trade policy imposed by the Chinese central government, we are able to differentiate between pilot cities and non-pilot cities at the city level, which creates a beneficial quasi-natural experiment, enabling us to use the difference-in-difference approach for empirical research.

The measurement of service productivity has been discussed as well. Productivity is typically measured by the total factor productivity (TFP) and labor productivity [40]. Previous studies have evaluated the TFP of the service industry in China at both the provincial level [41] and at the industry level [15,42]. Owing to the difficulty of obtaining data, firm-level TFP is difficult to estimate in some microstudies; thus, labor productivity is often utilized in its stead [43,44].

3. Service Trade Liberalization Policy and Development of the Service Industry

In this study, we take advantage of the database of World Bank Enterprise Surveys and use service sales and employees to measure firm-level service labor productivity. Referring to their approach, we manually download the municipal total service value added and service employees from China Urban Statistical Yearbook [45] to calculate service labor productivity. This measure, as far as we know, is applied for the first time to service studies at the city level in China.

3.1. The Policies of China’s Service Trade Liberalization

In the last century, China has employed a protective trade policy in its service industries, thereby maintaining a low level of openness. However, since joining the WTO in 2001, the Chinese government has strived to promote trade liberalization and service trade negotiations. In response to the liberalization of service trade policies, foreign investors have been increasingly drawn to investing in China’s service sector, since China has devoted increased consideration to the liberalization of services in trade.

In this context, China has implemented various city-level plans to further liberalize the service trade. In 2015, the State Council of China issued the “Implementation Plan for the Innovation Pilot City of Trade in Services”, and consequently decided to launch a liberalization policy for the service industry in 19 cities, including Beijing, Shanghai, and Chongqing, in the following year. The Innovation Pilot City of Service Trade policy adopts cities as its unit of operation. Pilot cities are obligated to create local policy documents that reflect the spirit of the State Council’s Implementation Plan. The State Council of China expanded the pilot policy to encompass 28 cities in 2020. Second, the implementation of the Foreign Investment Law in 2020 is of great importance for broadening the scope of China’s service trade liberalization. Equally, the Foreign Investment Law is beneficial for encouraging the influx of foreign capital, safeguarding foreign capital, and regulating foreign capital management. Third, in order to foster the growth of cross-border service trade in the Hainan Free Trade Port, the negative lists for cross-border service trade in 2021 have been implemented. The negative lists will be progressively implemented throughout the nation.

This paper examines China’s service trade liberalization using a policy named innovation pilot city. This policy has been in place since 2016, with each pilot city carrying out numerous policy measures in practice. Taking Beijing as an example, the People’s Government of Beijing Municipality issued the “Comprehensive Pilot Implementation Plan” to advance the liberalization of the service trade, thus furthering the development of local services.

Pilot cities chosen for the liberalization of the service trade are located across China, including cities along the eastern and southern coasts, two cities in the central region, and four cities in the western region. Four cities have implemented the innovation pilot city policy of the service industry in special districts, namely, the Nanjing Jiangbei New Area, the Chongqing Liangjiang New Area, the Guiyang Gui’an New Area, and the Xi’an Xixian New Area, in combination with urban construction policies and service trade liberalization.

Table 1. Implementation time of service trade innovation pilot city policy.

City	Time
Beijing	2015.5
Tianjin	2016.2
Shanghai	2016.2
Haikou	2016.2
Sanya	2016.2
Danzhou	2016.2
Sansa	2016.2
County of Hainan	2016.2
Shenzhen	2016.2
Guangzhou	2016.2
Hangzhou	2016.2
Wuhan	2016.2
Chengdu	2016.2
Suzhou	2016.2
Weihai	2016.2
Harbin	2016.2
Nanjing	2016.2
Chongqing	2016.2
Guiyang	2016.2
Xi’an	2016.2

Source: Research report on innovative development of China’s service trade (2019) [46].

shows the time points when each city initiated the service trade innovation pilot city policy.

The innovative pilot city policies implemented by chosen cities are composed of two major parts. First, they include the spirit of service trade liberalization conveyed by the central government. Cities should take proactive steps to implement challenging projects in service trade and investment in order to enhance their appeal to foreign investors and create specialized mechanisms and methods for the liberalization of the service trade. Second, given the current progress of local services, local authorities should leverage their distinct features and implement unique strategies. Local government departments must provide assistance to the service industry in order to promote its growth, while service industry associations and the media are encouraged to monitor the progress of the industry. For example, upon the commencement of the pilot city policy for service trade liberalization in Hangzhou, two objectives were set, namely, to expand the scope of the service trade and to establish the reputation of the “Hangzhou service”. Leveraging the advantages of Hangzhou’s local services, such as digital, cultural, and tourism services, the brand effect seeks to unearth the potential of local transport services and intellectual property services and establish public services in the areas of finance, exhibition, health, and education. Over the five-year period following the implementation of the policy, Hangzhou’s service trade witnessed an annual growth rate of more than six percent. By 2020, Hangzhou achieved an export proportion of 70% in digital services, forming a pattern of development that has made it the leading digital trade city in China. Therefore, the implementation of a pilot city policy of service industry liberalization in Hangzhou has proven to be highly successful.

3.2. The Development of China’s Service Industry

Since China’s inclusion in the World Trade Organization (WTO), the service industry has experienced a significant growth. Over the past two decades, from 1999 to 2019, the value of the service industry has seen a steady rise, with the total added value increasing from CNY 3493.5 billion to CNY 53,423.3 billion (in nominal GDP), as shown in Figure 1. The annual growth rate of GDP reached 14.6%. The share of services in the gross domestic product has grown from 38.5% to 53.9%. During the same period, the number of jobs in the service industry increased from 192.05 million to 367.21 million, making up 47.4% of the total employment nationwide. The service industry’s advancement has a significant impact on broadening the GDP and encouraging employment.

According to the Australian economist Ross Garnaut, the growth rate of China’s economy over the past two decades is considered a miracle, particularly in the 21st century when developed nations are experiencing sluggish economic growth and slow productivity growth. This is especially impressive for a developing country with a population of 1.4 billion. The historical patterns of developed nations have indicated that the rise of a service economy is an inevitable outcome of achieving a specific phase of economic growth. The persistent increase in the proportion of the service sector implies that China’s economic structure is undergoing a transformation, indicating that the nation is transitioning into a predominantly service-oriented post-industrial society.

The service industry generally produces intangible products, which have short production chains, low intermediate inputs, and slow productivity growth. Due to the heavy reliance on labor input, it is difficult to improve production efficiency in the short term through the introduction of advanced technology, as in industrial production. The improvement of efficiency within the service industry itself is slow. However, the service industry can still contribute to the expansion of the industry value chain and the enhancement of production quality and efficiency in other industries through the development of transportation, finance, and research and development sectors (World Trade Report, 2019). In order for China to sustain rapid economic growth while adjusting its economic structure, it is imperative to enhance efficiency within the service industry.

This paper further documents the status of the productivity of China’s service industry. Few papers have discussed and reported the pattern of China’s service productivity, especially at the city level, rather than at the firm level or industry level [48]. This paper uses China’s city-level data to show an increasing trend of productivity within the service industry. We use the labor productivity of the service industry in each city to represent city-level service productivity. The labor productivity of each city can be determined by dividing the annual nominal value added of the service industry by the total number of jobs in the same industry, with the result expressed in CNY per person.

Here, we first show the city-level varying trends of service productivity in some representative Chinese cities. We select six typical cities from the 298 Chinese cities: Hangzhou, Wuhan, Xi’an, Shijiazhuang, Nanchang, and Nanning. All of them are the provincial capitals situated in the eastern, central, and western regions of China. Figure 2 illustrates the alterations in the service productivity of these cities. Generally, it appears that the service productivity of cities is on the rise, but different cities have experienced different degrees and speeds of growth over time. It shows that the cities have similar levels of service productivity in the initial year (2011), but obvious differences in terms of productivity occur in the final year (2019). The service productivity of Hangzhou and Wuhan is higher than that of other cities. For the speed of productivity growth, there is a subtle variation between the two time periods from 2011 to 2016 and from 2016 to 2019; the cities experienced a greater growth differentiation in the second period. Hangzhou and Wuhan maintained a high-speed productivity increase, but the others were more stagnant. From the dimension of regional differences, it is obvious that the cities located in the eastern and central parts of China are exhibiting higher productivity growth than those in the western cities.

In 2016, China initiated the policy of service trade liberalization by selecting several cities as pilot cities. The policy difference between pilot cities and non-pilot cities can help explain previous differences in productivity growth trends between the cities that we have found in Figure 2. Hangzhou and Wuhan are pilot cities that experienced high-speed productivity growth, but Xi’an, which is located in western China, and other non-pilot cities experienced slower productivity growth.

Furthermore, to compare the different trends between pilot cities and non-pilot cities picked by the innovation pilot city policy of service trade liberalization, we calculate each city’s annual growth rate of service productivity. Then, we calculate the average growth rate of the pilot cities and non-pilot cities in each year, which is shown in Figure 3. The data presented in Figure 3 demonstrate that the service productivity in pilot cities has grown at a faster rate than in non-pilot cities from 2015 onward, especially in 2016. The figure implies that a city’s trade liberalization policy in services has a positive effect on the city’s service productivity, which will be further proved in the following empirical studies. Though trade liberalization policy was implemented in 2016, the difference in productivity growth in 2015 indicates that those picked as pilot cities have expectations and moderate adjustments for the upcoming policy.

4. Empirical Strategy and Data Source

To investigate the effect of service trade liberalization on service productivity, this section will discuss and explain the empirical specification and data sources used.

4.1. Empirical Strategy

By evaluating the variations in service trade liberalization policies across cities as well as the before-and-after change, the impact of service trade liberalization on a city’s service labor productivity can be estimated using a difference-in-differences (DID) strategy. Referring to Topalova [49], we exploit two sources of differences in China to establish the DID setting: The time differences come from before and after a critical year in the implementation of the pilot policy of service trade liberalization, and the second difference, cross-city differences, are from pilot cities and non-pilot cities. The pilot cities are regarded as the treatment group in this paper, and the non-pilot cities are regarded as the control group. The identification relies on a comparison of outcome variables for the treatment group with those for the control group both before and after the implementation of the pilot city policy of service trade liberalization.

Nevertheless, we must take into account the possibility that certain time-varying city characteristics may be associated with both the outcome variable and the regressor, resulting in biased estimates. To address this, as Shi and Xu [50] suggest, we will incorporate two-way fixed effects into the empirical model. Additionally, it is possible that other policies implemented during the same period could influence the city’s service productivity, thus potentially leading to an overestimation of the true effects of the innovation pilot city policy. In light of this concern, we strictly controlled the years of the used data to eliminate the interference of other service trade liberalization policies.

This paper evaluates the effect of pilot city policy on labor productivity from 2006 to 2019 by employing the difference-in-difference (DID) methodology. The following linear regression has been estimated:

$$SL{P}_{ct}={\beta }_0+\delta Pos{t}_t\times Trea{t}_c+\gamma X_{ct}+d_t+d_c+{\epsilon }_{ct}$$

(1)

We exploit the variations over the city year. SLP_ct, which is the natural logarithm of the division of service value added and service jobs, is a dependent variable that represents the labor productivity of the city, c, in the year, t. Post_t × Treat_c is the dummy variable of the innovation pilot city policy, which is used as follows: Post_t was assigned a value of 0 for the years 2006 to 2015 and a value of 1 for the years 2016 to 2019. Treat_c is assigned a value of 0 if the city c has not been chosen as a pilot city for service trade liberalization or 1 if it has been selected. X_ct stands for the control variables that are specific to a city. By employing the DID strategy, it is possible to include city and year fixed effects; therefore, d_t and d_c are added to account for time and individual fixed effects. Finally, ε_ct is a random error with a mean of zero. To account for heterogeneity and serial correlations, we must calculate the standard error by clustering the data according to cities, thus accounting for potential correlations [51].

4.2. Data Source

1. 

Data of Explained Variable.

In accordance with Heshmati [40], this paper substituted the labor productivity of the service industry as a proxy indicator of service productivity (Productivity is typically measured by the total factor productivity (TFP) and labor productivity. Previous studies have evaluated the TFP of the service industry in China at the provincial level [41] and the industry level [15,42]. Owing to the difficulties of obtaining data, firm-level TFP is difficult to estimate in certain microstudies; thus, labor productivity is often utilized in its stead [43,44]. Moreover, Martino [52] demonstrates that labor productivity and TFP have a long-term and steady positive association).

The labor productivity of a city can be determined by taking the total added value of the service industry and dividing it by the total number of employees in the same industry. This paper draws upon city-level data from 2006 to 2019. This period’s samples can illustrate the productivity alterations before and after the pilot city policy and will not be impacted by any other service trade policies. The information regarding the service added value and service employment is sourced from the China Urban Statistical Yearbook [45]. Our dataset consists of 4451 observations from 298 prefecture-level cities in China between 2006 and 2019.

2. 

Data of Control Variables.

This paper will select control variables in accordance with the background of a pilot city policy for service trade liberalization, so as to ensure the randomness of the selection of pilot cities. The following factors will be considered while selecting the policy pilot for China: the level of local economic development and service industry foundation; local involvement in international trade and foreign capital inflow; the level of local human capital; and the degree of digitalization [53,54].

Therefore, the following control variables have been selected: (1) The economic scale (logarithm of nominal GDP, lngdp) and per capita income level (logarithm of nominal per capita GDP, lnpergdp) are used to reflect the level of local economic development. As economic development progresses, the requirement for services in the local area grows, making it imperative to enhance service productivity. (2) Government size (the proportion of government budget expenditure in nominal GDP, lngov) is used to reflect the role of the government’s macro-control in economic development. (3) The level of urbanization (the proportion of rural household registration in a city’s household registration system, urban) is used to reflect the degree of urbanization. (4) The level of actually utilized foreign capital (the logarithm of foreign investment, lnfdi), is used to reflect the achievement and potentiality of the urban utilization of foreign capital. (5) The level of human capital (the average years of education per capita, edu). When the advantage in human capital is more pronounced, the development level of the service industry is correspondingly higher. (6) The infrastructure of digitalization (the logarithm of internet broadband access ports, lninter) has an important impact on cross-border services. The control variable data are all from the China Urban Statistical Yearbook [45]. The statistical characteristics of the variables are shown in

Table 2. Statistical characteristics of the main variables at city level.

Variables (Abbreviation)	Observations	Mean	Std. Dev.	Min	Max
Service productivity ($lnslp$)	4451	−5.6547	6.4396	−8.0086	7.3541
Time-varying dummy ($Pos{t}_t$)	4451	0.31	0.46	0	1
Treatment dummy ($Trea{t}_c$)	4451	0.48	0.50	0	1
Policy dummy ($Pos{t}_t\times Trea{t}_c$)	4451	0.15	0.35	0	1
Economic scale ($lngdp$)	4451	12.4136	0.3387	10.533	13.032
Per capita income level ($lnpergdp$)	4451	4.7720	0.2381	3.8999	5.2154
Government size ($lngov$)	4451	0.1190	0.1202	0.0092	2.5210
Urbanization ($urban$)	4451	0.6332	0.3584	0.0735	3.5942
Actually utilized foreign capital ($lnfdi$)	4451	11.2499	0.5536	8.6021	12.284
Human capital ($edu$)	4451	6.1531	1.3797	2.8272	9.2371
Infrastructure of digitalization ($lninter$)	4451	7.1655	0.4238	4.9031	7.9919

Source: China Urban Statistical Yearbook (2006–2019) [45].

.

5. Results

5.1. Benchmark Results

This study evaluates the effect of China’s trade liberalization of the service industry on the city’s service productivity by analyzing the implementation of the pilot policy of the innovative development of the service trade. In order to evaluate the implementation effect of the pilot policy, the DID model (Formula (1)) was used and the regression results are displayed in

Table 3. Basic regression results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	lnslp	lnslp	lnslp	lnslp	lnslp	lnslp	lnslp	lnslp
Postt × Treatc	0.178 ***	0.162 ***	0.162 ***	0.161 ***	0.156 ***	0.154 ***	0.159 ***	0.154 ***
	(0.061)	(0.050)	(0.050)	(0.050)	(0.031)	(0.042)	(0.049)	(0.041)
lngdp		0.237 ***	0.452 ***	0.065 ***	0.253 ***	0.397 ***	0.266 ***	0.116 **
		(0.011)	(0.080)	(0.014)	(0.055)	(0.086)	(0.050)	(0.047)
lnpergdp			0.026	0.029	0.037	0.030	0.037	0.054 *
			(0.035)	(0.035)	(0.047)	(0.034)	(0.041)	(0.032)
lngov				0.130 ***	0.112 *	0.155 **	0.130 ***	0.111 *
				(0.033)	(0.061)	(0.073)	(0.042)	(0.060)
urban					0.130 ***	0.139 **	0.144 ***	0.079 **
					(0.043)	(0.069)	(0.042)	(0.040)
lnfdi						−0.02 ***	−0.013	−0.016 **
						(0.007)	(0.009)	(0.008)
lnedu							0.139 **	0.162 ***
							(0.069)	(0.009)
lninter								0.007
								(0.022)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	4451	4451	4451	4451	4451	4451	4451	4451
R2	0.673	0.679	0.683	0.727	0.735	0.790	0.802	0.808

Note: This table shows the baseline regression results. Standard errors are clustered at the firm level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

. Column (1) presents the estimated result without the inclusion of any other control variables; the coefficient of the policy variable is significantly positive, which indicates that the implementation of pilot policies for the innovation and development of the service trade in China is likely to be a key factor in improving the productivity of the service industry, which is in accordance with theoretical expectations.

The results from items (2) to (8) suggest that economic scale (lngdp), per capita income level (lnpergdp), government size (lngov), urbanization level (urban), and per capita years of education (edu) all have significantly positive impacts on the productivity of the city’s service industry, indicating that higher levels of economic development, increased per capita income, larger government expenditure, urbanization, and improved human capital are all able to contribute to improvements in the service industry.

The lnfdi coefficient is significantly negative, indicating that an increase in the scale of foreign investment has a detrimental effect on the productivity of the city’s service industry. It is possible that the introduction of foreign capital may lead to firms relying on foreign technology, which hinders independent innovation capabilities and, consequently, impedes the enhancement of service productivity. The coefficient of the control variable for the development of infrastructure in the service industry is not significant, indicating that its impact on service productivity is not apparent. As the regression results show, adding control variables influenced the coefficients of the interaction items, albeit only slightly. This suggests that while certain city-specific factors can influence the productivity of the service industry, service trade liberalization nonetheless helps to improve it. The results are strong evidence to show that service trade liberalization is of great significance to China, which supplements the findings of the World Trade Report and Lai and Chen [2,11].

5.2. Robustness Check

Parallel trend tests, placebo tests, and resetting the measurement of the main variables are used to assess the robustness of the DID model that is adopted in this paper.

5.2.1. Parallel Trend Testing

The difference-in-difference (DID) model assumes that in the absence of treatment, the difference between the treatment and control groups would be constant or ‘fixed’ over time. Assuming common trends that are conditional on the start of the trend (which means the same starting point for treated and controls) is practically identical to assuming that there are no confounding factors (i.e., that the matching assumptions hold) that are conditional on past outcomes. This can be represented geometrically in a linear modeling by ‘parallel trends’ in outcome levels between treatment and control groups in the absence of a treatment [14].

This paper utilizes the research method of Ariu et al. [55] to investigate the trends of change in the treatment group and the control group. This test is advantageous as it enables policy change to show a distinct course in the different regions. To investigate whether there are pre-existing differences between the treatment group and control group, we replaced the original interaction items with a set of dummies for the leading and lagging years. Those dummies were still interacted with by parts of the time-varying and cross-city dummy variables mentioned in the previous section, and we only changed the time-varying dummy Post_t into the dummy group post_t+k. By setting the year 2016 as the benchmark, we regressed these substitution interaction terms on the service output in order to find the time trends. The estimation specification of the pre-existing time trend is as follows, and the revised equation is shown as Formula (2):

$$SL{P}_{ct}={\beta }_k{{\displaystyle \sum }}_{k\ge -5}^{3+}Pos{t}_{2016+k}\times Trea{t}_c+\gamma X_c+{\alpha }_c+{\phi }_t+{\epsilon }_{ct}$$

(2)

In Formula (2), Post_2016+k is a yearly dummy variable with a value of 1 in the pilot year and a value of 0 in other years. The other variables are consistent with the DID Formula (1). This paper investigates the changes that occurred between 2011 and 2019, particularly concerning the implementation of the pilot service trade innovation policy in 2016, as demonstrated in Figure 4. After removing one base year, we empirically tested the random policy year, and all the regression results prior to 2016 were not significant, indicating that before the implementation of the service trade liberalization policy, there was no substantial difference between the treatment and control groups. The service productivity of the treatment group significantly improved after the pilot policy was put into effect in 2016, in comparison to the control group. Consequently, the samples fulfill the parallel trend test that is required for the double difference estimation.

5.2.2. The Placebo Test

The placebo test assesses overidentifying assumptions and makes the common trend assumption more plausible. However, the estimated effect of the treatment group in the DID methods could be biased for the following two reasons: First, the treatment is anticipated, and therefore, it has an effect even before it starts. This raises some concerns about when to measure covariates that are not supposed to be influenced by the treatment. Second, if anticipation could be ruled out, an additional test for the common trend assumption is needed, because any estimated non-zero effect would have to be interpreted as a selection bias and thus, would cast serious doubts on the validity of the identifying assumptions. The placebo test can help solve those concerns and become a specification test for common trend assumptions. Suppose that we have several pre-treatment periods, in this case, we could pretend that the treatment happened earlier and then measure the outcome after the pretend treatment but before the treatment actually happened.

Drawing on the existing research of Liu and Zhao (2015) [53], this paper employs a placebo test to examine the effect of the pilot policy by altering the implementation time. Assuming that the service trade liberalization pilot policy was implemented prior to 2016, we repeatedly run the empirical regressions to estimate whether there was still an impact effect of the service trade liberalization on a city’s service productivity. If the coefficient of the interaction term is significantly positive, it suggests that the augmentation of a city’s service productivity may be attributed to other policies or random factors. To ensure the robustness of our regression results, we set the policy impact times as 2009, 2010, 2011, 2012, 2013, and 2014, respectively. Columns 1 through 6 of

Table 4. Results of the placebo test at city level.

	(1)	(2)	(3)	(4)	(5)	(6)
	lnslp	lnslp	lnslp	lnslp	lnslp	lnslp
Postt × Treatc	0.054	0.062	0.068	0.067	0.070	0.097
	(0.057)	(0.055)	(0.055)	(0.056)	(0.059)	(0.065)
City level controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes	Yes	Yes
N	4451	4451	4451	4451	4451	4451
R2	0.673	0.679	0.683	0.727	0.735	0.790

Notes: This table shows the robustness results of placebo test. Standard errors are clustered at the firm level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

show the estimation of the interaction coefficients, assuming that the year of establishment of the pilot is the corresponding year of the six aforementioned years. The results indicate that the assumption of the pilot time is not significant, thus suggesting robust baseline results.

5.2.3. Alternative Measurement for Main Variables

1. 

Alternative measurements for productivity.

Service productivity is typically measured by the total factor productivity and labor productivity. This paper will use the total factor productivity of the service industry as an indicator of the urban service industry’s productivity. This paper utilizes the stochastic frontier approach to evaluate the total factor productivity of urban services, which is based on the literature of Wang and Hu (2012) [15]. The transcendental logarithmic production function is adopted as the form of the stochastic frontier production function.

$$\begin{array}{r}{lny}_{ct}={\beta }_0+{\beta }_{1}ln{k}_{ct}+{\beta }_{2}ln{l}_{ct}+{\beta }_{3}ln{k}_{ct}ln{l}_{ct}+{\beta }_4{\left(ln{k}_{ct}\right)}_2+{\beta }_5{\left(ln{l}_{ct}\right)}_2+{\beta }_6\left(ln{k}_{ct}\right)t\\ +{\beta }_7\left(ln{l}_{ct}\right)t+{\beta }_{8}t+{\beta }_9{t}_2+v_{ct}-u_{ct} \end{array}$$

(3)

ln_yct denotes the value added of the city’s service industry; ln_kct stands for the capital input; ln_lct indicates the labor input; c and t are cities and years, respectively; and v_ct represents random perturbations affecting production activity, with a mean of 0 and a variance of σ_v2. Additionally, u_ct denotes the technical loss term, which is independently distributed with a mean of μ and a variance of σ_u2. This paper uses the perpetual inventory method to estimate the capital input data in lieu of the unavailable capital stock data.

2. 

Using a constant price to measure the explained variables.

We have estimated labor productivity by city-level service value adds and employment, but the value adds we sourced are calculated by the current price. The current price does not exclude the impact of price changes on the change in added value. Rather, the value added calculated at constant prices eliminates the factors of price changes, and the comparison of different periods can reflect the speed of production. Productivity measured by constant prices can better reflect reality.

According to the China Statistical Yearbook, we can see the Consumer Price Index (CPI) for each year. China’s CPI is estimated by looking at nationwide prices of goods and services covering 262 categories in China, which is the best indicator to reflect changes in price. The CPI is calculated by dividing the price level of the current year by the price level of last year. So we can calculate the value adds measured at a constant price by dividing the current year’s value adds by the CPI. Then, we estimate service productivity using value added measured at a constant price and the number of service employees. SLP_const_ct is used to represent the service productivity measured at a constant price.

3. 

Alternative measurements for explanatory variables.

As foreign capital access is a crucial avenue for service trade openness, the registered capital of foreign-invested firms can be taken as a measurement of foreign capital access, thus acting as a proxy variable for service trade liberalization. Since the data of cities at the prefecture level and above are only available until 2016, the robustness test is confined to the 2006–2016 sample.

Table 5. Robustness tests with alternative variables.

	(1)	(2)	(3)	(4)	(5)
	TFP	TFP	lnslp_const	lnslp	lnslp
Postt × Treatc	0.102 ***	0.109 ***	0.132 ***
	(0.006)	(0.006)	(0.046)
lnfdi				0.047 ***	0.053 ***
				(0.007)	(0.007)
City level control variables	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes	Yes
N	4451	4451	4451	3476	3476
R2	0.697	0.706	0.799	0.670	0.682

Note: This table shows the robustness results of alternative variables. Standard errors are clustered at the firm level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

displays the results of the robustness test that were obtained after re-measuring the explanatory variable. First, columns (1) and (2) of Table 5 demonstrate the results of a robustness test to ascertain the effect of China’s service trade liberalization on China’s service productivity, which is measured by the total factor productivity of China’s service industry. The interaction term coefficients are still statistically significant at the 1% level. Secondly, column (3) estimated the impact of service trade liberalization on service productivity measured at a constant price, and the coefficient of the explanatory variable was still positive and significant. Secondly, columns (4) and (5) of Table 5 utilize the registered capital of foreign-invested firms as a proxy variable to measure the level of China’s service trade liberalization. The results also demonstrate that service trade liberalization has an impact on service productivity.

5.3. Heterogeneity Analysis

5.3.1. Regional Heterogeneity

Geographical location, economic basis, and policy differences have caused the eastern, central, and western regions of China to differ in terms of their economic development levels, opening degrees, effectiveness of policy implementation, and other aspects. Generally, the economic development level of the eastern coastal areas is higher than that of the central and western regions, leading to variations in the influence of service trade liberalization on the productivity of a city’s service industry in different regions. Therefore, this paper divides the sample cities into three types, namely, the eastern cities, central cities, and western cities, based on Whalley and Zhang’s [56] division of China’s regions. Subsequently, an empirical test was conducted on the basis of these three city locations. The regional distribution of the pilot cities is evenly spread across the eastern, central, and western regions of China, thus not influencing the selection of the control group and experimental group.

Table 6. Heterogeneity by regions and marketization.

	(1)	(2)	(3)	(4)	(5)
	Located Regions in China			Level of Marketization
	lnslp	lnslp	lnslp	lnslp	lnslp
	Eastern	Central	Western	High	Low
Postt × Treatc	0.092 ***	0.039 **	0.286	1.032 **	−0.0680
	(0.022)	(0.016)	(0.325)	(0.503)	(0.234)
City level control variables	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes	Yes
N	2085	1134	1232	1786	2147
R2	0.603	0.614	0.602	0.917	0.927

Note: This table shows the results of heterogenous tests. Standard errors are clustered at the firm level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

illustrates the consequences of China’s service trade liberalization on service productivity in the eastern, central, and western regions. The results of the interaction coefficient $Pos{t}_t\times Trea{t}_c$ of columns (1) and (2) were both significantly positive at the 1% and 5% levels, respectively, suggesting that the openness of the service industry has a positive impact on service productivity in the cities of eastern and central China. The cities of the eastern region have a more advantageous economic foundation and liberalization experience, which allows them to better capitalize on the opportunities brought by the liberalization of the service industry, thus driving improvements in service productivity. Meanwhile, the central region has experienced an accelerated industrial transformation in recent years, which has attracted investment from numerous multinational companies. This has resulted in an intensified competition, leading to an improvement in the service industry’s productivity. In comparison to the economic development level of cities in the eastern and central regions, the western region lags behind, making it difficult for the service industry’s opening policy to be quickly absorbed and benefit from the technology spillover effects of trade liberalization. The interaction coefficient reported in column (3) of Table 6 is not significant, which demonstrates that the service trade liberalization policy has no effect on improving service productivity in western China.

5.3.2. The Heterogeneity of Marketization Level

According to Beverelli (2017) [29], the effects of service liberalization on productivity are linked to the institutional quality of countries. Fan et al. [57] designed a marketization index to investigate institution quality at a regional level in China. The marketization index has five components: the commodity market, factor market, non-state (private) sector, market intermediaries, and the relationship between the government and the market. These dimensions enabled an extensive evaluation of the circulation of market resources and factors in various regions.

The marketization of a city can influence the implementation of trade liberalization policies. On the one hand, with a high degree of marketization, the circulation of resources and elements is convenient, and service trade liberalization is more likely to cause an expansion in the types of consumable services. On the other hand, the rise of marketization can improve the information asymmetry of the city’s markets, correct the price signals of the factor and product markets, reduce domestic trade barriers, and thus help firms reallocate resources. All these measures lead to greater productivity. Therefore, we suspect that the impact of the service trade liberalization policy on the labor productivity of the service industry in China may vary due to the disparities in the level of marketization across different cities.

We examine the heterogeneity of the impact of service trade liberalization policy at different levels of marketization in cities. Based on the marketization index of cities in 2019, cities are divided into two groups: those with a high marketization and those with a low marketization. Columns (4) and (5) of Table 6 illustrate the influence of the service industry’s openness on the labor productivity in cities with high and low degrees of marketization, respectively. It has been established that the labor productivity of the industry can only be improved by opening services in cities with high marketization. As the policy pilot is evenly distributed between cities with high and low degrees of marketization, this will not influence the selection of empirical samples. When the degree of marketization is high, service producers will benefit from the trade liberalization of the service trade and the innovation reform of the service industry, as it will reduce the costs of obtaining service intermediates from foreign and domestic sources, as well as the costs of end-product producers. In this way, regions with higher market orientations will be able to better take advantage of the positive effects of service sector opening [58].

6. Mechanisms

6.1. Theoretical Analysis

1. 

Technology spillover effect.

Trade liberalization can attract domestic and foreign capital to the opening industries, providing the possibility for technological progress. Firms can leverage their own R&D to create and provide higher value-added services. The opening up of services is often accompanied by the influx of foreign advanced service technology and management experience [16,59,60]. In addition, some services can more effectively harness the technology spillover effects of FDI and intermediate service imports if their own R&D capabilities reach a level that can further encourage firms to improve their productivity [61,62].

2. 

Competition effect.

In general, a few domestic service industries have a monopoly and could decide the prices of service products. However, trade liberalization has enabled foreign service suppliers to enter into a host country with their services and technologies, typically at a lower supply cost and higher diversity. According to Jensen et al. [63], service trade liberalization can encouraged new service providers to enter the market, thus intensifying the competition. The increase in the number of competitors and in the proportion of non-state-owned capital has a considerable effect on productivity, and the effect of competition is significant in China.

In summary, service trade liberalization can affect service productivity in three ways through the competition effect. Firstly, foreign suppliers with advanced technology can offer high-productivity services in the host country, causing the overall local productivity to increase. Second, domestic firms have to devote themselves to independent technological innovation to improve productivity and strive for competitive advantages in the face of intense foreign competition. Third, after service trade liberalization, the introduction of foreign technology will enable local firms to learn advanced technology and increase productivity.

3. 

Effect of human capital.

Trade liberalization pilot policy will bring welfare to local talent introduction, which will in turn draw more skilled labor to local service supplies, thereby improving the productivity of the local service industry. Owing to the intangibility and simultaneity of the production and consumption of services, services are intertwined with human capital. Baumol [64] noted that labor productivity in the service industry is not as high as that in the manufacturing industry, which is attributed to the difficulty in enhancing productivity related to labor skills in the service process. However, Cheng [65] argued that the development of producer services, such as finance and telecommunication, in the 21st century has enabled the provision of services over long distances, and has brought about a division of labor, which affects the demands for human capital. Additionally, improvements in human capital have created conditions for service productivity increases. Trade liberalization policies can facilitate the movement of skilled personnel to the region, promote professional education, and ultimately lead to increases in service productivity.

6.2. Empirical Test

6.2.1. Tests of Technology Spillover Effect

To examine whether service trade liberalization can result in productivity growth through the mechanism of technological spillover in China’s innovation pilot cities, we have developed an empirical model:

$$RDexpenditur{e}_{it}={\beta }_0+{\beta }_1·IP{C}_{it}+{\beta }_2·Contro{l}_{ijt}+{\omega }_i+{\delta }_t+{\epsilon }_{ijt}$$

(4)

The model incorporates RDexpenditure_it as an indicator of the city’s technological innovation ability. The city-level total expenditure of government R&D investment can be a measure of technological innovation capacity. After service trade liberalization, capital investment can be pushed towards a firm’s R&D, thus providing an opportunity for technological progress and a spillover effect on productivity. We have chosen a natural logarithm as the explanatory variable to investigate the effect of the trade liberalization policy. The data on government R&D expenditure are sourced from the China Urban Statistical Yearbook [45].

The data in column (1) of

Table 7. Mechanism test results.

	(1)	(2)	(3)
	RDexpenditure	Competition	High Skills
Postt × Treatc	0.149 ***	0.517 **	0.0670 *
	(0.0562)	(0.243)	(0.0389)
City level controls	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
City FE	Yes	Yes	Yes
N	3951	3946	3951
R2	0.936	0.788	0.947

Note: This table shows the results of mechanism discussions. Standard errors are clustered at the firm level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

reveal the effect of service trade liberalization on a city’s technological innovation capacity. We take into account fixed effects to alleviate the endogenous problems that may cause bias. Findings show that the coefficient of the policy variable is significantly positive at the 1% level, which reveals that the level of technological innovation in the service industry increases when the policy is implemented, suggesting that technological innovation could be a potential method of enhancing the productivity of the service industry. Our result, from the perspective of service trade liberalization, is a supplement to the view of the spillover effect of technology from trade in the literature of Bas [7], etc.

6.2.2. Tests of Competition Effect

To estimate the level of competition, this paper refers to the strategies of Duranton and Puga [66] and Shao [67] to formulate a city’s industry diversity index. Similarly to the designed Herfindahl (HHI) index, the reciprocal of the sum of the squares of the employment shares of the industries present in the city is calculated. In addition, 15 service industries are based on the categories of the China Standard Industry Classification. By calculating the proportion of the number of employed people in each industry to the number of employed people in the tertiary industry in each city, we can calculate the industry diversification index. The formula for the calculation is as follows:

$$Di{v}_{it}=1/{{\displaystyle \sum }}_{n=1}^N{(S_{int}/S_{it})}^2,N=15$$

(5)

The industry diversification level of city i in year t, referred to as Div_it, is determined by the ratio of the number of employees in the nth service industry to the total employment. The higher the index, the greater the competition in the city’s service industry. By combining the employment data by industry in the “China City Statistical Yearbook” over the years, it is possible to estimate the competitive effect of the service market.

Service trade liberalization policy has resulted in the emergence of new firms and new types of services in certain industries, leading to an increase in employment and a more diversified job market. As the value of S_int decreases, the industrial structure of the city will become more diversified. An increased diversification index indicates that competition in a city has become more intense. Firms will therefore take more active measures to raise productivity in order to enhance their domestic market competitiveness and survive the more intense market competition.

Thus, to test competition effects, we created an empirical model where Competition_it symbolizes the competition between cities, which is gauged by industrial diversification (DIV_it).

$$Competitio{n}_{it}={\beta }_0+{\beta }_1·IP{C}_{it}+{\beta }_2·Contro{l}_{ijt}+{\omega }_i+{\delta }_t+{\epsilon }_{ijt}$$

(6)

The second column in Table 7 reveals that the influence of service trade liberalization on service productivity is significantly positive at the 10% level. This result is consistent with the previous theoretical analysis—that the level of competition in the city could be a potential impact mechanism of service openness to improve productivity. The market competition effects of China’s service trade liberalization on services are similar to those found by Su and Shao [17], even though they mainly discuss manufacturing firms.

6.2.3. Tests of Human Capital Effect

The productivity of the services could be dependent upon the level of human capital, so we intended to analyze whether the service trade liberalization of the service industry could augment productivity by raising the human capital level in those innovation pilot cities. Due to the fact that high-skilled labor is often represented by the number of university graduates, city-level employment data can be used to estimate city-level human capital. Here, we are able to calculate the level of high-skilled labors in the service industry by multiplying the proportion of a city’s bachelor’s degree employment jobs with the total number of service industry employment jobs in each city.

We have formulated an empirical model in which Highskills_it denotes the number of tertiary industry graduates employed in year t in city i.

$$Highskill{s}_{it}={\beta }_0+{\beta }_1·IP{C}_{it}+{\beta }_2·Contro{l}_{ijt}+{\omega }_i+{\delta }_t+{\epsilon }_{ijt}$$

(7)

The results of column (3) in Table 7 show that the policy will significantly improve the level of human capital. This result demonstrates that improvements in human capital in each city are a potential mechanism for service trade liberalization to enhance productivity. This result supplements two branches of the recent literature. On the one hand, Li et al. [18] found that trade liberalization in China has long-term negative effects on the formation of human capital, but our findings show that service trade liberalization in China contributes to improvements in human capital. The reason for this, we guess, is that policy can bring a greater influx of high-skilled labors to pilot cities in a short time. On the other hand, compared with U.S. service markets that suffer from trade shocks and bear high unemployment [68,69], this paper shows how the service market in the biggest developing country, China, is benefiting from trade liberalization, and improving overall human capital and efficiency.

7. Conclusions

7.1. Conclusions and Policy Implications

This paper utilizes the data from 2006 to 2019 at the city level and constructs a DID model to examine the effect of China’s service liberalization on the productivity of the service industry, as demonstrated by the quasi-natural experiment of the service trade innovation pilot policy. The city-level service labor productivity is estimated by the division of the city’s service value adds and service employment, and the pattern of productivity growth was shown in this paper as well. Additionally, by exploiting the cross-city, cross-time variation at the time of China’s innovation pilot city policy of service trade liberalization, we employ a difference-in-difference strategy to estimate the effects on service productivity. The results of the study are as follows:

First, by liberalizing the service industry, China has improved the productivity of the service industry. This paper verifies the robustness using the parallel trend and placebo tests. Simultaneously, this paper also substitutes the explanatory and explained variables with alternative variables, and the results are still robust. Moreover, it has been found through heterogeneity analysis that the effect of liberalizing the service industry on productivity is significantly impacted by regional differences and the level of marketization development. Cities in the eastern and central regions are better positioned to take advantage of the service trade liberalization policies, leading to an increase in the productivity of the service industry. The western region is lagging behind in terms of opening the service industry. In regard to the degree of marketization, service openness can enhance the efficacy of the service industry in cities with a high level of marketization, yet it has no influence on the productivity of the service industry in areas with a lower degree of marketization. Furthermore, this paper further examines and verifies the potential mechanism of the service industry’s liberalization to enhance its productivity and concludes that technology spillover, competition, and human capital are the three channels that have been invoiced to achieve the goal in question.

China’s evidence highlights the positive relationship between trade liberalization and efficiency in the service industry, and it strongly emphasizes the benefits of service trade liberalization. The following policy implications are inspired by the findings of this paper:

The service trade in China should be gradually liberalized and trade restrictions should be diminished through service trade liberalization policies and other approaches. This appeal applies not only to China but also to the vast number of developing countries. As developing countries are often at a disadvantage in product competition, developing the service industry is a short cut to growing economies. They should capitalize on the advantages of an open economy, maximize the advantages of trade liberalization, and achieve economic growth. Precisely, the government should also pay attention to the institutions of the service industry and the service trade. This includes the construction of a sound legal system, improvements in the service industry’s information technology, the development of a modern logistics system, and so on. In addition, the government should be mindful of the disparities in economic development and marketization between different regions and adjust the policy to the local contexts; otherwise, the benefits of trade liberalization will be relatively insignificant. Upholding the concept of trade-led growth, developing countries are expected to reap greater rewards and sustainable development from trade liberalization in the long run.

7.2. Limitations and Future Potentials

Research on the relationship between service trade liberalization and service productivity is an important topic for sustainable economic development and it needs further study. Perspectives of firm-level and city-industry level estimations of service productivity, as well as more precise measurements of productivity, can be considered. In addition, in the process of service trade liberalization layout, how to coordinate the implementation of the service trade agreement signed by the central government and the implementation of service industry policies of trade liberalization by local governments still needs further discussion. All of the aforementioned limitations could potentially become the focus of future research investigations.