Can the county to district reform in China restrain corporate earnings management? From the perspective of regional local protection

ABSTRACT Under the strategic background of regional coordinated development, the County to District can remove administrative barriers and reduce local protection. Based on the matching data of Chinese A-share listed companies from 2000 to 2018 and the County to District Reforms, this paper empirically explores the influence of County to District Reforms on earnings management. We find the County to District Reform can effectively reduce earnings management, especially in areas with high local protection. Further mechanism analysis shows the County to District Reforms reduce local protection by increasing credit availability for enterprises, and ultimately restrain earnings management. Additional analysis shows the County to District Reforms increase the bank loans of enterprises in areas with higher local protection. This paper deepens the understanding of the County to District Reforms and enriches the literature on how government actions can affect corporate governance.


Introduction
Implementing the regional coordinated development strategy and correctly handling the relationship between the government and the market have always been one of the major goals of the Chinese government. In November 2018, the Central Committee of the Communist Party of China and the State Council issued the Opinions on Establishing a New Mechanism for More Effective Regional Coordinated Development, which is a reflection of this strategic goal. Specifically, the County to District Reform in China, which arose in the 1980s, is also an administrative division adjustment to develop the regional economy, straighten hierarchical relations, and exert government functions. As part of China's governance, the County to District Reform can improve the relationship between local governments and the market, promote resource allocation efficiency, and optimise the business environment of enterprises (Peng et al., 2019). Then, can the County to District Reform affect corporate governance of micro-enterprises? From a micro perspective, earnings management is an important aspect of corporate governance. It internally affects the information transparency of enterprises and externally affects the market information environment. Vicious earnings management behaviour also seriously damages the interests of investors and reduces capital market allocation efficiency, which hinders capital market development (Biddle et al., 2009;Francis et al., 2009;Zhou & Chen, 2008). However, most studies focus on the effects of County District Reforms at the macro level. Few scholars have focused on corporate governance (Tang & Wang, 2015;Wang & Nie, 2010;Zhan & Zeng, 2021).
To explore how the County to District Reforms affect corporate governance, we examine the effect on earnings management. It should be pointed out the Fiscal Reform of the Province-Manage County is another administrative division adjustment in China that aims at promoting the financial level of reformed areas. Specifically, it refers to changing the financial relationship from the 'province-city-county' into 'province-city' and 'province-county', which means the financial transactions between provinces and counties are settled directly with counties instead of prefecture-level cities. However, the original intention of the County to District Reform is to achieve a higher level of market integration and regional coordinated development. Prior literature shows that the decentralisation of autonomy is an important reason for local protection. Compared with districts, county-level governments have higher autonomy in finance and administration, which causes serious local protection (Han, 2020;Xu, 2011). After County to District Reforms, powers are re-centralised to prefecture-level governments, which can eliminate administrative barriers and reduce local protection. Therefore, based on local protection, we examine how the County to District Reform affects earnings management behaviour.
Taking Chinese A-share listed companies from 2000 to 2018 as the sample, we empirically explore the effects of County to District Reforms on earnings management. The results show that (1) the County to District Reform can significantly reduce enterprise earnings management, (2) mechanism analysis shows that the County to District Reform can restrain earnings management by reducing local protection and refining channel to increase bank loan availability for enterprises, (3) cross-sectional test indicates that when enterprises belong to manufacturing industry or face serious financing constraints, the improvement effect of County to District Reforms on earnings management is more remarkable, and (4) further analysis shows that after the County to District Reform, bank loans obtained by enterprises evidently increase.
The contributions of this paper are reflected as follows. First, it enriches the literature on the policy effects of County to District Reform. As mentioned above, most scholars have investigated the consequences of the County to District Reform from a macro perspective (Tang & Wang, 2015;Wang & Nie, 2010;Zhan & Zeng, 2021). Based on corporate governance, we explore the effects of County to District Reform on earnings management, deepening the understanding of County to District Reform.
Second, based on government favouritism, Lu and Chen (2017) find the County to District Reform reduces government subsidies received by enterprises but does not consider the lending behaviour of financial market. Correspondingly, based on the market lending behaviour, we find the County to Districts Reform breaks administrative barriers between counties and districts, increases credit availability for enterprises, and helps ease financing constraints, which ultimately reduces earnings management. Therefore, this paper complements prior studies (Lu & Chen, 2017).
Third, we also expand the studies on county governance and corporate governance. From the perspective of tax efforts, G.Z. Li and Jia (2019) find the Fiscal Reform of the Province-Managed County can reduce earnings management by enhancing the tax efforts of county-level governments. Based on local protection, we find the County to District Reform can restrain earnings management by reducing local protection and increasing access to credit resources for enterprises.
Finally, previous literature shows that governments intervene in the credit resources allocation of financial institutions, which reduces resource allocation efficiency (K. Li & Ye, 2007;Yu & Pan, 2008;Zhou, 2004). We find that by reducing administrative barriers and local protection, the County to District Reforms improves the credit allocation efficiency of financial institutions, which has certain practical guiding significance.
The rest is organised as follows. Section 2 details the institutional background, literature review, and hypotheses development. We describe data and variables in Section 3. Empirical results are presented in Section 4. Section 5 conducts additional analysis. Section 6 concludes this research.

Institution background
The County to District Reform refers to prefecture-level cities (or direct-administered municipalities) changing their counties into districts. According to the opinions of most scholars, motivations of County to District Reforms include the following aspects: (1) Expanding the development space of central cities and breaking through development bottleneck.
(2) Alleviating the regional market segmentation caused by administrative barriers and achieving regional coordinated development (Hao et al., 2021). (3) Solving historical issues, such as 'one city and one district', and optimising the administrative structure of national land space. (4) For political promotion, the prefecture-level government also favours the County to District Reform. The reason is that higher-level governments examine the performance of lower-level governments based on economic performance. By conducting County to District Reforms, county-level resources are concentrated in prefecture-level cities and promote urban economic development (Zhan & Zeng, 2021). Additionally, the removal of counties into districts is also an adjustment to meet the requirements of improving urbanisation and administrative management. Districts adopt urban management and counties adopt rural management. In some counties, urbanisation has reached a high level and are inadaptable to rural management continuously.
The County to District Reforms also need to meet certain conditions. Main policy documents are the Report on Adjusting the Standards for the Establishment of Cities formulated by the Ministry of Civil Affairs in 1993 and the Standards for the Establishment of Districts formulated in 2003 and 2014. The above documents stipulate standards for newly established districts in population size, industrial structure, economic aggregate, and fiscal revenue. For example, as stated in the Standards for the Establishment of Districts (2014), counties that intend to establish districts should be included in the central city planning and the urbanisation rate should not be less than 50%.
For counties and districts within the same prefecture-level city, there are evident differences in geographic location, affiliation, resource acquisition, economic incentives, and management authority. Considering the geographical location, districts are usually located within the scope of the central city and included in the development planning of the central city. However, counties are geographically far from central city and do not have such development advantages. As for affiliation, districts are directly managed by the prefecture-level governments, while counties have higher independence and decision-making autonomy, such as independent fiscal responsibilities and executive powers (Ji & Zou, 2019). Compared with counties, districts have more economic resources. Specifically, we manually sort the average number of financial institutions in counties and districts separately from 2000 to 2018. As shown in Figure 1, the average number of financial institutions in counties is less than that in districts, which indicates that compared with enterprises in districts, the loans that county-level enterprises can obtain from local banks are relatively scarce.
Some regional restrictions in the development of the Chinese financial market can also be observed, and these restrictions cause severe regional market segmentation, making it difficult for county-level enterprises to obtain cross-regional loans. As reported by Chinanews.com in 2012, in a regional economy, the fund demand and supply among counties are different. However, the central bank's loan is allocated directly to each county and stipulates it cannot be adjusted among counties, which generates the resources mismatch and restricts coordinated development. 1 Similarly, the China Banking and Insurance Regulatory Commission issued the Opinions on Promoting Rural Commercial Banks to Adhere to Positioning, Strengthen Governance and Improve Financial Service Capabilities in 2019, which emphasises that rural commercial banks should serve the local area and the new loanable funds should be used mainly for the local area in principle. Trying to reduce the effects of administrative barriers and local protectionism on factor markets, the Central Committee of the CPC and the State Council issued the Opinions on Accelerating the Construction of a Unified National Market in 2022, which states that local protection and regional barriers should be firmly removed, and no region can engage in a self-centred cycle of 'small but complete', let alone regional blockade in the name of 'internal circulation'.

County to district reform
From the macro-level, scholars have made explorations, including how the County to District Reforms affect economic growth, local public finance, population, and urbanisation (Tang & Wang, 2015;Wang & Nie, 2010;Zhan & Zeng, 2021). For example, Wang and Nie (2010) point out that administrative divisions directly affect regional economic development. Further, Zhan and Zeng (2021) find that the agglomeration effect of County to District Reforms through the expansion of urban scale can improve the quality of economic development in local cities. Some scholars examine the influence on population and urbanisation development and draw a consistent conclusion that the County to District Reform is conducive to population growth and alleviate urbanisation imbalance (Tang & Wang, 2015). Based on the micro-level, fewer scholars have conducted related research. Fan and Zhao (2020) document that the County to District Reforms reduce the enterprises' tax burden and promote regional economic growth. Taking Chinese industrial enterprises as the sample, Lu and Chen (2017) find that County to District Reforms reduce government subsidies and tax incentives for enterprises within the jurisdiction. While Peng et al. (2019) show that although County to District Reforms reduce subsidies and financing preferences, they strengthen product market competition and inhibit zombie enterprises.

Local protection and credit intervention
Under the Chinese-style decentralisation, a phenomenon cannot be ignored is the serious administrative barriers and local protection between different administrative regions, namely, local governments intervene, restrict, and exclude inter-regional economics for the sake of local economic interests, eventually leading to market segmentation and hindering market integration (Bai et al., 2004;H. Li & Zhou, 2005;Young, 2000;Zhou, 2004). To boost economic growth and safeguard the interests of various economic entities, local governments are particularly inclined to intervene in the local credit allocation, hoping to obtain more credit resources for the local area and restrict credit resources outflow (He et al., 2008;K. Li & Ye, 2007). Further, they also guide the local credit resources flowing to industries or projects supported by governments. Undoubtedly, local protection causes severe financial market segmentation, leading to disorderly market competition and reducing resource allocation efficiency.
Most studies on County to District Reforms are from the macro perspective. Using Chinese industrial enterprises' sample, some scholars investigate whether the County to District Reforms affects decision-making of micro-enterprises. As a way of county governance, few scholars explore the effects on corporate governance. Taking Chinese A-share listed companies as the sample and based on the local protection, we empirically test whether the County to District Reforms affects the earnings management behaviour of enterprises.

County to district reform and earnings management
Earnings management motivations include capital market motivations, contracting motivations (such as lending contracts and management compensation contracts), and regulatory motivations (Fields et al., 2001;Hearly & Wahlen, 1999). We make our exploration based on lending contracts. Usually, as the debtor, the enterprise determines the creditor-debtor relationship by signing a lending contract with the creditor. Although the accounting information in debt contracts can reflect the debtor's solvency, because of inherent limitations of information, debt contracts are easily affected by earnings management (Lu et al., 2008).
As for debt financing, before lending contracts are signed, enterprises need to be evaluated by banks, bond rating agencies, or public market regulators. The evaluation criteria are based mostly on financial condition, mortgage guarantee, and profitability of enterprises (Petersen & Rajan, 1994). However, corporate financing is often limited because of market frictions, such as information asymmetry and principal-agent problems (Stiglitz & Weiss, 1981). Previous studies suggest that to obtain credit resources and reduce financing costs, enterprises would conduct earnings management to beautify financial statements (Cassar et al., 2015;DeFond & Jiambalvo, 1994;Ma et al., 2014;Sweeney, 1994). Moreover, Chinese commercial banks cannot identify earnings management when providing loans and even issue more long-term loans to enterprises with serious earnings management (Lu et al., 2008). 2 Therefore, under the lending contracts, enterprises have incentives to conduct earnings management to obtain scarce credit resources. We also advocate that compared with enterprises in districts, county-level enterprises face more limited credit resources and have stronger earnings management motivations.
First, influenced by Chinese fiscal decentralisation and political promotion tournament, local governments are actively committed to developing the local economy, which gives birth to local protectionism (H. Li & Zhou, 2005;Zhou, 2004). To support local enterprises and protect them from the market competition, local governments often intervene in the credit decision-making of financial institutions and place more credit resources in local enterprises, which distorts the resource allocation efficiency in financial market and leads to serious financial market segmentation (K. Li & Ye, 2007;Yu & Pan, 2008). Moreover, local protectionism often exists in governments of multiple levels (Hao et al., 2021).
The County to District Reform belongs to the upward transfer of management power. After the county is restructured into a district, most administrative and economic management authorities of original county-level governments are transferred to prefecture-level governments. The independence and autonomy of county-level governments have been seriously reduced, which weakens their local protection motives and results in a higher level of regional governance and business environment. First, it can alleviate regional financial market segmentation and promote the credit resources flow across regions. Second, original counties will be managed directly by prefecture-level governments, such as districts, integrated with the central urban area, and enjoy preferential policies of prefecture-level governments after the reform, which can reduce government frictions between administrative regions.
Second, the distribution of financial institutions in Chinese is uneven. Specifically, the number of financial institutions in counties is much less than that in districts and the former is dominated by rural credit cooperatives and rural commercial banks. As pointed out by Cai et al. (2020), the administrative barriers of jurisdiction and branch system of Chinese commercial banks hinder the crossregional flow of credit resources, and thus, the location between banks and enterprises is an important factor affecting enterprises' access to credit resources. 3 By weakening administrative boundaries, the County to District Reforms promote regional market integration, which is conducive to the cross-regional flow of credit resources and original county-level enterprises can obtain more loans from other banks in prefecture-level cities to broaden financing channels.
Based on the above analysis, we expect the County to District Reforms can reduce local protection and increase the credit resources of enterprises, to reduce earnings management by weakening the lending contracts motivation. Therefore, we raise the hypothesis: H1: The County to District Reform can restrain the earnings management behaviour of enterprises within the jurisdiction.

County to district reform, local protectionism and earnings management
The Chinese hierarchical structure includes four levels: central, provincial-level, prefecture-level, and county-level governments. Governments of different levels have different administrative and economic management authorities. Consequently, this administrative decentralisation leads to local protectionism and market segmentation.
Before the County to District Reform, county-level governments have higher independence, stronger local protectionism, and more severe market segmentation between counties and districts. Under the circumstances, companies in counties could perform business activities within the local area, which makes it difficult for them to enter the market of districts. Accordingly, access to credit resources and other production factors are more restricted. After the reform, local protection declined (Han, 2020). At the macro-level, original administrative barriers are broken, which helps the prefectural-level governments to make unified decisions on the overall planning and industrial layout and promote market integration (Tang, 2019). At the micro-level, the County to District Reforms reduce local protection and increase access to credit resources for enterprises in local area, to ultimately alleviate lending contracts motivation and reduce earnings management. Further, we formulate a second hypothesis: H2: In regions with higher local protection, the County to District Reform has a more significant improvement effect on earnings management.

Data sources
We take Chinese A-share listed companies from 2000 to 2018 as the sample to explore the effects of County to District Reforms on earnings management. We obtain information on County to District Reforms from the website of the Ministry of Civil Affairs of China. Financial data, registered address information of enterprises and distribution information of financial institutions are from CSMAR database. Government subsidy data and market index are from Wind database. China's Annual Industrial Survey data are used to calculate EG index (Ellison & Glaeser, 1997).
The sample processing is as follows: (1) Identify the province, prefecture-level city, county, or district where firms are located according to the registered address information and match the corresponding administrative division code (specific to the county level) for each firm.
(2) Given that Beijing, Tianjin, Shanghai, and Tibet have great differences in administration, enterprises registered in the above regions are dropped (Lu & Chen, 2017). (3)  After the above sample selection, we obtained 24,213 firm-year level observations. To eliminate the effect of outliers, we winsorise all continuous variables at the upper and lower 1% levels.

Earnings management (DA1)
Following Kothari et al. (2005) and G.Z. Li and Jia (2019), we adopt the cross-sectional modified-Jones model with performance-matching to measure earnings management (DA1 i;t ). First, we use the relevant financial data of enterprises to perform regression by industry and year employing model (1) and obtain coefficients (α 0 , α 1 , α 2 , α 3 , α 4 ). Next, we substitute the above coefficients into model (2) and calculate the nondiscretionary accruals (NDA i;t ). Finally, we estimate discretionary accruals (DA1 i;t ) by model (3).
In the following models, TA i;t represents total accruals, which equals to NI i;t minus CFO i;t . NI i;t denotes the net profit of listed firm i in period t and CFO i;t represents the net cash flow generated from operating activities. A i;tÀ 1 is the lagged total asset. ΔREV i;t denotes changes in operating income. ΔREC i;t represents changes in accounts receivable. PPE i;t is a net fixed asset. ROA i;tÀ 1 denotes the lagged return on assets.

County to District Reforms (Treat � Post
The core explanatory variable is Treat � Post. Treat denotes whether enterprises are affected by the County to District Reform. If enterprises are affected by the reform during sample period, Treat equals 1 and 0 otherwise. Post denotes whether the reform is implemented, which equals 1 if the year is after the County to District Reform and 0 for other cases.

Control variables
We choose size, age, lev, soe, growth, loss, roa, bdsize, inp, and audit as control variables. Appendix A shows the specific definitions of variables.

Baseline regression
Logically, we predict that the County to District Reform contributed to restraining the earnings management behaviour of enterprises. Following Beck et al. (2010), we conduct an analysis using the time-varying differences-in-differences model as follows: In specification (4), DA1 i;t denotes earnings management of firm i in period t, Treat � Post denotes the County to District Reform, and Control i;t is the series of control variables, which are listed in Section 3.2. μ i is the firm fixed effect, τ t is year fixed effect, and ε i;t is the residual term. Considering the differences between enterprises in different counties and industries, we further control county fixed effect and industry fixed effect, and cluster the robust standard errors by county-level in all regressions.
We focus on the coefficient of Treat � Post. If β 1 is significantly negative, it supports our hypothesis that the County to District Reform can reduce earnings management. Table 2 reports the results. The coefficient of Treat � Post is significantly negative at the 5% level, suggesting that the County to District Reform can restrain the earnings management behaviour of enterprises within the jurisdiction, which verifies H1. From an economic perspective, the coefficient of Treat*Post is −0.010, indicating that the enterprises' earnings management level in the original counties decreased by 0.14 standard deviations (0.01/0.07) after County to District Reforms.

Parallel trend and dynamic effect tests
Using DID analysis requires meeting the common trend assumption, that is, before the reform, the change trend of earnings management in the treatment and control groups should be consistent without a systematic difference. To check the parallel trend assumption and dynamic effect test, following Beck et al. (2010) and G.Z. Li and Jia (2019), we conduct specification (5). where Treat denotes whether enterprises are affected by County to District Reforms, current means the treatment year of the County to District Reform, Pre2 refers to year 2 and beyond before the treatment year. Pre1 refers to one year before the treatment year. Post1 indicates one year after the treatment year. Post2 denotes year 2 and beyond after the treatment year. The list of control variables and fixed effects are the same as in specification (4). Figure 2 depicts the result and shows that the estimations of Treat � Pre2, Treat � Pre1; and Treat � current are not significantly different from zero, thereby suggesting before the County to District Reform, no obvious difference between the treatment and control groups could be observed, meeting parallel trend assumption. Meanwhile, the estimation of Treat � Post1 and Treat � Post2 is negative and significant at the 1% level, indicating the earnings management reduces significantly after County to District Reforms.
We perform PSM-DID analysis to overcome systemic differences caused by the County to District Reform and endogeneity of sample selection. We take enterprises affected by the County to District Reform from 2000 to 2018 as the treatment group and those not affected by the Reform are taken as control group. First, we match the control group for enterprises that have experienced the County to District Reform. The matching method is as follows: (1) To ensure similarity of characteristics, the control and treatment groups are restricted to the same prefecture-level city. (2) We take the control variables in baseline regression as covariables, perform Logit regression, and obtain propensity scores. (3) Using 1:3 proximity matching, we match control group with treatment group. (4) We conduct a balance test. The result shows that our treatment and control groups could pass the balance test.
We then use the matched sample and perform regression analysis. As shown in Column (1) of Panel A, Table 3, the coefficient of Treat � Post is significantly negative, suggesting that our results are robust.
(2) Excluding samples that are in districts all the time Considering before the County to District Reform, quite a difference between companies in districts and companies in counties can be observed. Following Hao et al. (2021), we eliminate enterprises that are in districts all the time. Column (2) of Panel A in Table 3 reports the result, which further supports our hypothesis. (

3) Considering interference policy
When the County to District Reform gradually rises in China, the Fiscal Reforms of Province-Manage County are carried out simultaneously. Could the effect of the County to District Reform on earnings management be interfered with by the Fiscal Reforms of a Province-Managed County? Referring to Fan and Zhao (2020), we add the variable of Province-Managed County Reform (SZG) and conduct regression following specification (4). As shown in Column (2) of Panel A, the coefficient of Treat � Post is negative, which means the improvement effect on earnings management will not be affected by the Fiscal Reforms in the Province-Managed County.
Finally, we conduct other robustness tests. 4 (1) To avoid measurement deviation, we use the adjusted Jones model, DD model, and McNichols model to measure earnings management (Dechow & Dichev, 2002, 2002DeFond & Jiambalvo, 1994;McNichols, 2002). (2) We check our conclusion by taking a sample of Chinese industrial enterprises. The untabulated results show that our conclusion is robust. (3) We divide the whole sample into private enterprises, central state-owned enterprises, and local state-owned enterprises. The unreported result presents that the County to District Reform has a more remarkable improvement effect on earnings management in local state-owned enterprises.

County to district reform, local protection, and earnings management
In this section, we explore the underlying mechanism. Similar to the above analysis, we expect the County to District Reforms to restrain earnings management by reducing local protection. We conduct a grouping regression to verify the mechanism. If the  improvement effect on earnings management is more significant in the sample with high local protection, H2 is supported. First, we measure local protection based on administrative boundary segmentation. Under the Chinese decentralisation system, serious market segmentation among different administrative regions can be observed. The county-level governments have independent financial and administrative power, and thus, they are motivated to protect local enterprises (Hao et al., 2021;Young, 2000). Accordingly, we construct two indicators to measure local protection: (1) the number of surrounding counties, defined as the number of counties within 10 km from the border of a county and (2) the number of counties in a particular prefecture-level city plus 1. The more 'surrounding counties' or 'counties in the particular prefecture-level city', the more serious the local protection. Then, based on the average level of local protection, we divide the full sample into two groups (low and high local protection groups) and regress, respectively, by specification (4).
Panel A of Table 4 reports the results. As shown in Columns (2) and (4), the coefficient of Treat � Post is merely significant in the 'high local protection group', indicating that the County to District Reform could restrain earnings management behaviour by reducing local protection. Thus, H2 is verified.
Second, we use the EG index (Ellison & Glaeser, 1997) to measure local protection. The EG index is usually used to measure industrial agglomeration and local protection will reduce industrial agglomeration. Therefore, the EG index can reflect the degree of local protection, and a higher EG index means lower local protection (Bai et al., 2004). The calculations are as follows: Subscript i represents industry, c represents county, and t represents year. G cit is the Gini index, calculated as G cit ¼ P c x ct À s cit ð Þ 2 . x ct is the proportion of staff in county c to the whole country, and s cit is the proportion of staffs of industry i in county c to industry i. H cit is the Herfindahl index, where Z cit is the proportion of single enterprise staff to industry i of county c. γ cit represents EG index. Furthermore, based on the administrative division code and year identification, we match enterprise data with EG index. Then, we calculate the median value of the EG index by year and divide the sample into the low (EG index higher than the median value) and the high local protection groups (EG index lower than the median value). Because of data limitations, the sample period in this test is 2000-2013.
Columns (1)-(2) of Panel B in Table 4 reports the grouping regression results. The table shows that in the subsample with high local protection, namely Column (2), the coefficient of Treat � Post is −0.049 and significant at 1%. While in the subsample with low local protection, namely, Column (1), the Treat � Post is negative but not statistically significant. Thus, we draw a consistent conclusion based on the EG index (1997).
Finally, we use the marketisation index, which is compiled by Fan et al. to measure the marketisation of each province as a proxy for local protection. A higher marketisation means lower local protection. Columns (3)-(4) of Panel B in Table 4 reports the results. It   Continued) also suggests that the effect of the County to District Reform on earnings management is merely significant in the high local protection sample.

Further mechanism analysis from the perspective of local loan availability
The above analysis shows the County to District Reform can improve earnings management by reducing local protection. However, what benefits can be brought to enterprises by decreasing local protection? Logically, we believe the County to District Reform can reduce local protection and further increase the local loan availability, which finally reduces earnings management. To test this channel, we adopt the cross-grouping regression. First, we measure local protection by the number of surrounding counties and divide it into two subsamples: the low and the high local protection groups (the first group). Next, we divide the two subsamples into four subsamples by the local credit availability (the second group). If the improvement effect on earnings management is only evident in the subsample with high local protection and low credit availability, it can verify that the reform restrains earnings management behaviour by reducing local protection and increasing the local credit availability. Referring to Cai et al. (2020), we use the number of banks in the prefecture-level city where the enterprise is located to measure access to the local credit resources. 5 Before cross-grouping, considering the lower local credit availability and the higher the local protection, we divide the entire sample into two groups based on local credit availability and perform regressions, respectively. As shown in Panel A of Table 5, the coefficient of Treat � Post is merely remarkable in the 'low local credit availability group', which supports H2.
We then perform the cross-grouping regressions mentioned earlier. The results are presented in Panel B of Table 5. We find that the effect of County to District Reform reducing earnings management is only significant in the subsample with high local protection and low credit availability, which is in line with our expectations.

Effect of county to district reform on enterprise loan
After County to District Reforms, we determine whether bank loans obtained by enterprises have increased. Specifically, we construct specification (6) to explore the Presents the results of mechanism analysis. In Panel A, we take administrative boundary division as proxies for local protection. Column (1)-(2) use the number of surrounding counties. Column (3)-(4) use the number of counties in a particular prefecture-level city. In Panel B, we use EG index in Column (1)-(2) and marketisation index in Column (3)-(4) to measure local protection. *, **, *** denotes significance at the 10%, 5%, and 1% level, respectively. t statistics in parentheses.
relationship between the County to District Reform and enterprise loans. As for bank loans, we refer to Zhang and Li (2012) and define it as the proportion of long-term loans to total loans. The higher the proportion and the more preference the enterprises have to obtain bank loans.
Similarly, the number of surrounding counties is used to measure local protection and Table 6 reports the results. Panel A is the result of grouping regression based on local protection. As Column (2) shows, the coefficient of Treat � Post is 0.091, which is significant at the 1% level, meaning that for the region with high local protection, after the County to District Reform, long-term bank loans obtained by enterprises will increase by 0.31 standard deviation (0.091/0.29). Panel B is the result of the cross-grouping regression. It suggests that the positive effect of County on District on enterprise loans is more evident in the high local protection and low credit availability group, which is also consistent with mechanism analysis.   Shows further mechanism analysis. Panel A is grouping based on local credit availability and Panel B is cross-grouping of credit availability and local protection. *, **, *** denotes significance at the 10%, 5%, and 1% level, respectively. t statistics in parentheses.

Cross-sectional analysis
(1) Industry Attributes The manufacturing industry usually has a high level of profits and taxes, which has become an important protection industry for local governments. Intuitively, we advocate that the effect of the County to District Reform on earnings management may be more evident in manufacturing industries. To test our prediction, we divide the entire sample into two groups: the non-manufacturing and the manufacturing industry groups. As presented in Columns (1)-(2) of Table 7, the coefficient of Treat � Post is negative in the 'Manufacturing industry group'.
(2) Financing Constraints An important reason that enterprises conduct earnings manipulation is financing constraints. Accordingly, we predict that the more serious the financing constraints faced by   enterprises, the more County to District Reform can alleviate the financing constraints to a greater extent by reducing local protection and increasing the credit resources available to enterprises, which can further reduce earnings management. Therefore, we conduct the cross-sectional analysis based on financing constraints. If the above conjecture is true, the improvement effect on earnings management will be more significant in the sample with high financing constraints. Following Kaplan and Zingales (1997), we use KZ index to measure the financing constraints of enterprises and divide the sample into two groups with low and high financing constraints. Columns (3)-(4) of Table 7 present the results. As expected, the improvement effect of the County to District Reform on earnings management is more obvious in enterprises with high financing constraints.

Alternative explanations
Based on the market-oriented lending behaviour of financial institutions, we document that after the County to District Reform, the local protection decreases, administrative barriers are removed, and enterprises can obtain credit resources more easily. However, from the perspective of government behaviours, the increase in enterprise credit resources may be due to government support, that is, after the original county becomes a district, it gains more attention from prefecture-level governments, which usually takes the form of government subsidies and tax incentives. We make the following explorations to exclude alternative explanations.
First, we use the ratio of government subsidies to total operating income as a proxy for government support and divide the entire sample into two groups, one with high government subsidies and the other with low government subsidies. If the increase in credit resources is caused by government actions, the relationship between County to District Reform and earnings management will be more significant in the sample with low government subsidies. As shown in Columns (1)-(2) of Table 8, the coefficient of Treat � Post is only significant in the group with high government subsidy, which is contrary to the above expectation, thereby suggesting that the increase in the enterprise credit resources is not the government action but rather market lending behaviour.  Shows the cross-sectional analysis. *, **, *** denotes significance at the 10%, 5%, and 1% level, respectively. t statistics in parentheses.
Second, we check whether the improvement in earnings management is caused by a change in tax efforts. G.Z. Li and Jia (2019) find that the Fiscal Reform of the Province-Managed County enhances the tax efforts of local governments and restrains earnings management. Fan and Zhao (2020) find that the County to District Reforms weaken the tax effort of county-level governments and reduce tax burden of enterprises. Based on these studies, we expect that with the decrease in local government's tax efforts, the County to District Reform would aggravate earnings management, especially for enterprises with authority from the local government to collect taxes and for enterprises with a higher tax burden. However, this expectation contradicts our finding, suggesting that our conclusion is not caused by weakening tax efforts. To test it further, we make explorations in two ways. First, we use the corporate income tax scaled by the total asset to measure enterprises' tax burden and divide the sample into low taxes group (tax burden lower than average level) and high taxes group (tax burden higher than average level). Then, we conduct grouping regressions. As shown in Columns (3)-(4) of Table 8, the coefficient of Treat � Post is not significantly positive in the 'high taxes group', indicating that the County to District Reforms do not increase the earnings management of enterprises with a high tax burden. Second, by referring to G.Z. Li and Jia (2019), we divide the samples into two groups based on the tax collection authority of local governments. That is, if an enterprise was founded earlier than 2002, it indicates that the local government has the authority. If it was founded later than 2002, it means the local government has no authority. The results are presented in Columns (5)-(6) of Table 8. We find Treat � Post is significantly negative rather than positive in the 'Earlier than 2002 group'. The above tests demonstrate that our conclusion cannot be explained by weakening tax efforts.

Does trans-regional operation affect the policy effect of county to district reform?
More enterprises have adopted the trans-regional operation with the improvement of marketisation. Enterprises can obtain various geological resources by establishing subsidiaries across regions, which is helpful in alleviating financing constraints of single region operation. The policy effect on such enterprises may be weakened if enterprises operate across regions.  Presents the results of ruling out the alternative explanation of supported by governments. *, **, *** denotes significance at the 10%, 5%, and 1% level, respectively. t statistics in parentheses.
Therefore, we attempt to examine whether the trans-regional operation affects the relationship between the County to District Reform and earnings management. Compared with single companies, group companies are more likely to operate transregionally. Hence, we divide the sample into group and single companies. In prior studies, the definition of enterprise group is as follows: take the listed company as the starting point and look for the final controlling party through the control chain. If the final controlling party owns other entities in addition to the superior company, then the listed company belongs to an enterprise group and the value is 1, otherwise, it is 0 (W.J. Li & Yan, 2021). We also refer to the above study to define group companies. However, because of the limitations of the data of direct controlling shareholders in the CSMAR database, the sample period in this test is 2003-2018. Table 9 shows that the relationship between County to District Reforms and earnings management is only significant in single companies, indicating that the transregional operation of enterprises would weaken the policy effect of the County to District Reform.

Conclusion
Since the reform and opening up, administrative division adjustments at the county level in China have occurred frequently. However, studies on policy effects of the County to District Reform are mostly based on the macro-level. Hence, we examine the effects of County to District Reforms on corporate earnings management behaviours from the perspective of micro-enterprises.
Taking Chinese A-share listed companies from 2000 to 2018 as the sample, we find that (1) after the County to District Reforms, the degree of earnings management is significantly reduced.
(2) Mechanism analysis shows that the policy of the County to District restrains earnings management by reducing local protection. In particular, we document that after the local protection decreases, local loan availability increases, which weakens their motivation for earnings management, causing the earnings management to decrease accordingly. (3) We further conduct a cross-section analysis based on the industry attributes and financing constraints. The results indicate that compared with non-manufacturing companies, the Table 9. Does trans-regional operation affect the policy effect of county to district reform?.
County to District Reform has a more significant improvement effect on the earnings management of manufacturing companies. This effect is more obvious when enterprises face greater financing constraints. (4) We also investigate the relationship between the County to District Reforms and enterprise loans. The results indicate that the bank loans obtained by enterprises evidently increased after the County to District Reforms. Our conclusions have certain significance for the government and enterprises, which reveal that the county governance policies at the macro-level can also affect the corporate governance of micro-enterprises.