1. Introduction
Since its political transition in the 1980s, China has gradually moved from a planned economy to a market economy. However, it is undeniable that the economic development of cities in China is still closely related to the influence of administrative power, which the local government can flexibly rely on to intervene in the economy. Administrative power has an obvious spatial hierarchy, which can be reflected in the administrative level of the city [
1]. In China, the central government has the highest administrative and economic power, which decreases as the administrative level decreases: municipalities have the same social and economic management power as provincial administrative regions, then prefecture-level cities, and finally county-level cities. Cities at different administrative levels have different economic management powers, political influence, and resource allocation capabilities [
2], which not only have important implications for macro-level economic development but also for enterprises.
The development of enterprises is affected by many factors; a large number of studies showed that the total factor productivity (hereinafter referred to as TFP) is the most important factor affecting the sustainable development of enterprises [
3]. However, firstly, previous studies mainly focused on the influence of changes happening on specific city levels on enterprise TFP, thus belonging to static analysis; secondly, the existing literature mainly focused on the short-term impact, and few studies examined the impact on enterprises’ long-term TFP from administrative-level promotions. The changes in cities’ administrative levels can be seen as giving the local government more autonomy; although they are still bound to the central government, the promoted cities have fewer restrictions than before and the officials have much more power in the economic development.
Chongqing has long been part of Sichuan Province. Although it became a separate state-planned city in 1983, it still belonged to Sichuan Province administratively. In 1997, Chongqing separated from Sichuan Province and was elevated from a prefecture-level city to a provincial-level city with the same administrative powers as Sichuan Province. Therefore, Chongqing’s upgrading provides us with a unique sample to study the impact of city-level upgrading on firms’ TFP.
First, the new border between Chongqing and Sichuan was an unexpected and exogenous change. Although the demarcation of Chongqing was discussed before 1997, the exact boundary of Chongqing was not announced until it was approved by the central government on 14 March 1997. In fact, the central government had considered four different plans for dividing Sichuan, but it was unclear which plan would be implemented until the central government made a final decision [
4]. Second, as mentioned above, under China’s unique political system, the decision-making power of local governments depends heavily on their position at the political level. Therefore, the upgrading of Chongqing allowed the newly established municipality to gain substantial powers in economic, administrative and personnel matters. Third, in many cases where new administrative divisions are established, there is usually a significant change in the ethnic composition [
5]. In contrast, both in Chongqing and Sichuan, the Han population accounts for around 93% of the total population, and the elevation of Chongqing has not led to significant changes in ethnic composition, so we can rule out the potential impact of changes in ethnic composition. Finally, Chongqing has been part of Sichuan Province for a long time, so it has similar characteristics to Sichuan in terms of culture, system, and social development path.
Taking Chongqing’s upgrading to a municipality at the prefecture level in 1997 as an exogenous shock and industrial firms in 2013 as a research sample, we investigate the long-run effects of city-level upgrading on firms’ TFP using the spatial regression discontinuity approach. The contributions of this paper are as follows: first, the existing research on the relationship between political hierarchy and TFP mostly focuses on the effect of the existing city level on TFP, which is a static effect, while our paper focuses on the long-term effect of city-level elevation on firms’ TFP, which is a dynamic effect. Second, by using Chongqing’s elevation as an exogenous shock and the Sichuan-Chongqing border as a dividing line, the spatial regression discontinuity approach can effectively identify the causal relationship between the administrative level and TFP. Moreover, we also analyze the mechanisms involved, which not only helps us to understand the problem but also provides a factual basis for how to solve the problem. Third, the existing literature on city level and TFP studies has mostly focused on short-term effects, but as Krugman stated, productivity is noteverything, but in the long run, it is almost everything [
6]. Therefore, this paper provides an important addition and useful extension to the existing literature by showing the long-term effect.
This paper is organized as follows:
Section 2 provides the institutional background.
Section 3 provides the theoretical analysis.
Section 4 explains the research data and identification strategy.
Section 5 presents the analysis and discussion of the results.
Section 6 reports the mechanism analysis.
Section 7 concludes the paper.
3. Theoretical Analysis
In a unitary state system such as China, the central government has the highest administrative power, which has an obvious spatial hierarchy [
1]. As the administrative level decreases, the corresponding administrative power decreases. The central government relies on administrative power to regulate and control the economy. The rank of the administrative level largely determines the amount of political and economic resources, which largely affects the flow and redistribution of resources. Therefore, the upgrading means that Chongqing has gained greater economic and social management authority, and can more flexibly use government intervention to develop the economy, because of which officials can obtain a greater probability of promotion under China’s unique political system [
8]. At the same time, changes in land management systems have resulted in land playing an increasingly important role in local economic development [
11]. Under the current land system, the local government has a monopoly over land supply and can control land prices; to promote industrialization and economic development, the government tends to sell industrial land at low prices and build industrial parks to attract enterprises to settle [
12]. Since the land management authority is concentrated in the central and provincial governments [
13], Chongqing became autonomous in land management authority after it was separated from Sichuan province.
Based on this point, it can be inferred that raising the administrative level can affect firm TFP in the following two ways: on the one hand, the expansion of land management authority allows the Chongqing government to deliberately lower the land transfer price to attract more firms to settle, which leads to an abundance of many low-productivity firms and the “crowding out” of high-productivity firms, as a result, firm TFP will decline. Many studies used firm-level microdata to confirm the negative impact of low land transfer price on firm TFP [
14,
15,
16]. The other aspect is that under the political centralization and economic decentralization system in China, the Chongqing government has a stronger incentive to develop the economy after the city’s administrative level is upgraded, which will promote the government’s policy intervention, which will lead to the misallocation of firm resources and lower firm TFP. Several scholars used empirical tests to support the above view [
17,
18]. Under the unique development mode of regional competition and promotion of competition in China, land has become an important tool for local governments to attract investment and promote local economic development. In order to gain competitive advantages, local governments often adopt the “race to the bottom” regional competition strategy in the process of attracting investment, and even sell land at a price far lower than the land expropriation cost or even promote “zero land price” to accelerate the process of local industrialization and achieve the goal of economic development. After the elevation of Chongqing, local officials have a stronger incentive to sell land at a lower price, which will largely distort the land transfer price. For enterprises, the low land price not only reduces the one-time prepaid capital of enterprises, alleviating the liquidity constraint during the enterprise’s founding period, but the industrial land acquired through the agreement transfer also becomes its assets, and the industrial land can be mortgaged to banks for financing the development of enterprises. Therefore, low land prices may have attracted more companies with poor prospects and excess capacity, dragging down TFP.
4. Research Data and Identification Strategy
The data used in this paper are industrial enterprise data from 2013, industrial census data from1995 and land market concession data from 2013, among which the industrial enterprise data are pre-processed by referring to Nie et al. and Chen [
19,
20]. We restrict the data sample to Sichuan Province and Chongqing; the distribution of industrial enterprises is shown in
Figure 1. We can see that, first, the distribution of firms near the center of Chengdu and Chongqing is relatively dense, while the western part of Sichuan province and the eastern part of Chongqing are less dense. Second, many industrial firms are distributed near the border of Sichuan and Chongqing, which is a prerequisite for using the spatial regression discontinuity approach in this paper.
The explanatory variable in this paper is firm TFP. The methods commonly used in the literature to calculate firm TFP include OLS, FE, OP, and LP, among which OP and LP are considered to be more efficient than OLS [
21].We use OLS to calculate TFP for the following reasons: first, the sample of industrial firms in 2013 can only use the OLS method to calculate TFP because of the lack of key indicators when using OP, LP, and ACF methods, such as industrial value added, intermediate goods input. Secondly, the main disadvantage of the OLS method is that it cannot solve the simultaneity deviation and the sample selection deviation. However, fortunately, the existing literature suggests that the TFPs obtained by different methods are not significantly different [
22]. Third, the existing literature such as Hsieh and Klenow’s study also uses OLS to calculate TFP [
23]. Therefore, we use the Solow residual method to calculate firm-level TFP. To check the robustness of the results, we also use labor productivity as an alternative measure of TFP.
Chongqing has long been part of Sichuan Province, and the border between Chongqing and Sichuan Province was only an intra-provincial border until 1997, which ensured that the bordering countries had similar cultural and institutional environments as well as geographic and economic conditions before the shock. This provides us with a very good condition to use the spatial regression discontinuity approach. Therefore, we can use the borders as the discontinuity to conduct a comparative study of firms. The regression has the following form:
is the firm-level total factor productivity (TFP), and
is a dummy variable defined as 1 if the firm is located in Chongqing and 0 otherwise.
is a polynomial controlling geographic location to ensure that the function is smooth at the boundary. Following Dell [
24], we use a two-dimensional polynomial in the longitude and latitude of the firm, which can absorb any smoothing tendency in the boundary results.
is the coefficient we are interested in, indicating the effect of Chongqing’s administrative-level elevating on the firm’s TFP.
is the independent and identically distributed error term. Considering that firms in different industries of different counties may be correlated, we use the standard errors of clustering at the county-industry level.
As discussed by Gelman and Imbens [
25], Equation (1) can be estimated using two methods: the nonparametric local linear regression or the global polynomial regression. In our benchmark regression, we used local linear regression, which uses a narrow bandwidth near the boundary and controls for a linear latitude–longitude polynomial. As there is no widely accepted two-dimensional optimal bandwidth [
26], we restricted our sample to firms within 30 km of the border (
Figure 2). We also consider bandwidths of 10, 20 and 50 km to ensure that our estimates are robust to a particular choice of bandwidth. Finally, to test the robustness of the results, we report results using a global multidimensional regression.
Table 1 shows the t-test results of the main variables for the counties within 30 km of the Chongqing–Sichuan border. It shows that the counties near the border do not differ significantly in slope and altitude, so they can be considered as smoothed. The same applies to the average annual temperature. The data are from meteorological stations in 2013. As complete climate data at the county level were not available, we use the data from the weather stations to calculate the climate data for each county, following the method of Shephard [
27]. Specifically, the distance between the county and each weather station is used in the construction of the weights, i.e., the meteorological characteristics of a county should be more like the closer weather station. The weight of each weather station is constructed as follows:
represents the weight of the weather station, and
represents the spherical great circle distance between the weather station
i and county
c. We calculated the distances from the border counties to downtown Chengdu and downtown Chongqing, where the local government is located. The results show that there is no significant difference in the distance from the border counties to Chengdu, while there is only a difference at the 10% significance level to Chongqing. Overall, we can conclude that there is no significant difference in the geographical characteristics of the counties on the border between Sichuan and Chongqing.
Given that macroeconomics has an important impact on the development of enterprises, we also examine the economic characteristics of border counties. The GDP, GDP per capita, gross industrial output value per capita and urbanization rate are obtained from the statistical yearbooks of Sichuan Province and Chongqing Municipality in 1996. The GDP, GDP per capita and gross industrial output value per capita of the counties in Chongqing are all slightly better than those of Sichuan Province, but the urbanization rate is slightly lower than that of Sichuan. Importantly, none of the
t-test results are significant, indicating that there is no significant difference in the economic characteristics of the border counties before Chongqing’s elevation. We also compare the share of manufacturing and the share of ethnic minorities in the border counties with data from the 2000 Chinese Census, and the results are still not significantly different. Finally, we examine the TFP of firms located 30 km from the border between Chongqing and the rest of Sichuan Province using 1995 Census data.
Figure 3 shows that there is no significant difference in the TFP of industrial enterprises located on either side of the border line in 1995, and the coefficients in
Table 2 support this conclusion. The above results show that there is no significant initial economic and industrial difference between the counties near the border, i.e., the conditions for using the spatial regression discontinuity approach are met.
7. Conclusions
Under China’s current administrative system, political hierarchy measures the amount of political and economic resources that the local government can control, directly affecting the flow and redistribution of resources, and it thus has important implications for firm behavior. This paper empirically examines the long-term effects of city-level upgrading on micro-firm TFP, using Chongqing’s upgrading to a provincial-level municipality as an exogenous shock and data on industrial firms in Chongqing and Sichuan in 1995 and 2013. The results show that: (1) there is no significant difference in TFP between firms near the border of Chongqing and Sichuan before Chongqing’s separation, but after Chongqing’s elevation to a provincial-level municipality, the TFP of Chongqing firms is significantly lower than that of Sichuan provincial firms, by about 5.5–7.4%. (2) In the robustness test, we used the DID regression discontinuity approach, replaced the TFP measure, control for fixed effects, adjusted the bandwidth and boundaries, used placebo tests, and eliminated the effect of firm migration. All test results are supportive of our conclusions. (3) The mechanism analysis finds that after Chongqing’s upgrading, the land price in Chongqing is 25% lower than that in Sichuan Province, which means that under the incentive of economic development, Chongqing attracts more low-productivity firms to settle, leading to the “crowding out” of high-productivity firms. At the same time, we find that Chongqing’s upgrading exacerbates the degree of resource misallocation of firms through policy intervention. Together, land price and resource misallocation lead to a decline in firms’ TFP.
The findings of this paper provide empirical evidence for the debate on the relationship between the government and the market. Under China’s unique political system, the elevation of the city level has expanded the government’s economic and social management authority, and thus has a significant impact on firms. China has claimed to promote a better combination of an efficient market and a responsive government. Accordingly, this paper suggests that the efficiency of land use should be improved to avoid inefficient government investment; policies such as the establishment of a land inspection system should be explored. Second, government intervention should be appropriately reduced, and ex-ante review and ex-post monitoring of government intervention should be strengthened.
Although the process of administrative elevation is rather exclusive to countries with very strictly top-down political structure such as China, this paper still provides significant insight into the role the government can play in the economic development process. The effect the administrative power can have on the private enterprises is largely dependent on the incentive structures for the ones wielding such power. In a truly representative democracy, because the prosperity of the private enterprises is of great importance to the local residents, thus empowering the local government can be rather beneficial, whereas in an authoritarian government structure, the officials are incentivized by the desires of pleasing their higher-ups who picked them for their jobs, whose priorities do not always align with the that of the local private enterprises. The empirical evidence provided by this paper can be used against the idea of giving the government more power in economic management, when the representativity of such government is questionable.