Empirical Analysis of China’s Agricultural Total Factor Productivity and the Reform of “County Administrated by Province”: Insights from Agricultural Enterprise Data

Abstract

The study focuses on examining how the total factor productivity (TFP) of China’s agriculture-related enterprises is affected by the reform of “County Administrated by Province” (CAP). Using panel data from 1998 to 2013, with a sample size of 292, 423 agriculture-related enterprises, the study investigates the influence of CAP reform on the TFP. The findings reveal a significant dampening effect of the CAP reform on the TFP of agriculture-related enterprises. These results are further supported with a series of robustness tests including placebo test, multidimensional fixed effects test, and clustering of cities or counties. Heterogeneity analysis reveals that the CAP reform has significantly restrained the TFP of agriculture-related enterprises in high level cities, while showing no significant impact on cities with lower administrative levels. In the eastern region, the impact of the CAP reform on the TFP of agriculture-related enterprises is significantly negative, whereas it is not statistically significant in the central and western regions. the study on mechanisms elucidates that the inhibitory impact of CAP policy on the TFP of agriculture-related enterprises is enhanced by urban property prices, labor resource allocation, and banking competition. The research conclusion is of significance in guiding the practices of agriculture-related enterprises and deepening the reform of “County Administrated by Province”.

1. Introduction

Agricultural enterprises are the backbone of rural revitalization in developing economies, and enhancing the productivity of agricultural enterprises through technological innovation is the only way to achieve the high-quality development of rural industries. Agricultural enterprises, as a micro-entity of rural economic development and a major source of income for farmers, play an indispensable role in rural revitalization. However, most of the current research focuses on the total factor production ratio of agriculture. Total factor productivity improvement in agriculture is mainly constrained by three factors: technological change, technological efficiency, and public policy [1,2,3]. And it has been shown that technical efficiency has a limited effect on productivity improvements during the period analyzed. In contrast, technological change has a more positive impact on production. This has been confirmed in studies in the EU, the Czech Republic, and China [4,5,6].

The institutional environment is one of the key factors influencing the long-term impact of firms’ TFP, and a favorable institutional environment can help to increase total factor productivity in manufacturing. Reform of administrative management systems can often drive technological progress and promote total factor productivity in micro-enterprises [7]. Studies have found that the policy environment has a significant impact on total factor productivity in agriculture in transition and developed economies, For example, African countries’ trade policy reforms have increased agricultural TFP, while agricultural subsidies in China and rural tax reforms have both significantly increased agricultural TFP. In addition, for Poland [8] and Italy [9] and the EU [10,11], empirical studies in developed economies such as these also confirm the ability of public policies to reduce regional differences in TFP. And in studies of the above literature, the above empirical evidence suggests that governments can enhance the productivity of agro-related enterprises through administrative reforms, thereby contributing to rural revitalization.

Unlike the state and county or provincial county administration systems that are widely implemented worldwide, China has generally implemented a “province-city-county” management system. In the early 2000s, the central government enacted a fiscal reform known as the Provincial-Management-County (PMC) fiscal reform. The Chinese government began reforming the direct province-county system in 2002. Five years later, 24 states have implemented direct province-county reforms, except for a few minority autonomous regions that have not adapted to the reforms. Forty percent of the counties and municipalities have carried out pilot reforms of the financial system, and more than two hundred counties and municipalities have carried out pilot projects to strengthen the county level and expand local management authority. On 9 July 2009, the Ministry of Finance issued a document on the reforms of direct province-counties, with the exception of the few minority autonomous regions that have not adapted to the reforms. provincial counties to carry out reforms. That is, by the end of 2010, all provinces, with the exception of ethnic minority areas, will have carried out direct financial reforms at the county level. The reform of counties under direct provincial control transformed the existing hierarchical financial management system from a model of provincial and municipal management, and county and city management, to a model of direct provincial management of county, city, and county financial affairs alone. Its core elements are the division of revenues and expenditures, transfer payments, fund flows, financial budgets and accounts, and year-end settlements [12]. The main purpose of the “provincial directly administered counties” reform is to improve administrative efficiency and reduce the financial pressure on county governments by eliminating prefecture-level cities as an intermediate layer between provinces and counties [13], which consists of two main models: fiscal decentralization and economic decentralization. The central government has been successful in promoting economic growth by gradually decentralizing socio-economic decision-making and management to local governments and easing the fiscal pressure on county governments. However, whether PMC’s decentralization reforms lead to increased inter-regional spending and widen regional spending inequalities will also depend largely on the variability of local governments in terms of their spending capacity and structure [14].

In 2022, the Chinese government proposed an ambitious overall development strategy of “rural revitalization”, one of the core objectives of which is to achieve high-quality development of rural industries by enhancing the efficiency of enterprises through scientific and technological innovation. In order to provide historical experience for the “rural revitalization” strategy, a question that deserves to be explored in depth is: Has China’s “direct control of counties” reform improved the productivity of agriculture-related enterprises? In order to answer this question, the thesis will use micro-level data on agricultural enterprises to conduct an in-depth analysis. The findings of the study will help to clarify the specific impact of the “direct provincial control” reform on the productivity of agricultural enterprises at the micro level, which has important implications for further reform of the local administrative system to promote rural revitalization and provides useful lessons for other economies. The remainder of the paper is organized as follows: Part II presents a review of relevant studies; Part III presents the hypotheses of the paper; Part IV introduces the data sources, variable definitions, measurement methods, and econometric models used in the paper; Part V presents descriptive statistics and robustness tests on the empirical results, as well as heterogeneity analysis and endogeneity tests, and finally, analyzes the specific impact mechanisms; Part VI presents Conclusion and Implications.

2. Literature Review

2.1. Research on the Policy of “County Administrated by Province” and Similar Problems

The policy of “county administrated by province” is a kind of administrative management reform of local governments. The research literature in this field can be summarized from such three aspects as promotion of economic growth, activities performed by local governments, and financial problem-solving at the grassroots level.

First, the county reform reduces management hierarchies and endows county-level government with greater decision-making powers in fiscal and tax policies in terms of the promotion of economic growth [15]. The reform of “county administrated by province” eliminates the potential exploitation in the middle link, directly governs the county and endows it with more expenditure liabilities, and distributes the power of withholding fiscal transfer payment of the provincial government to the county government. And the decentralization of government power will have a significant impact on economic growth [16]. In high-income countries in Western Europe, fiscal decentralization has a significant impact on economic growth [17]. The AR model research, however, found that the fiscal decentralization is beneficial to the economic growth of Nigeria for the long term [18]. According to regression analysis and the finite element method, the reform of “county administrated by province” can significantly promote the growth of per capita fiscal transfer payment at the county level. However, Li et al. [19] put forward different views on fiscal investment, land transfer, and information exchange that the reform of “county administrated by province” inhibits the per capita GDP growth at the county level, and the county government is managed directly by the province to expand the management scope of provincial governments, and, thus, the land corruption occurred frequently due to increasing difficulty in supervision and communication.

Secondly, in terms of activities of local governments, governments around the world actively implement power decentralization to strengthen governance and improve policy efficiency [20,21]. According to research on data from 590 counties in China from 2000 to 2009, the reform of “county administrated by province” drove the county government to employ more active fiscal policies, and become more active in political competition of economic growth [22]. On the other hand, based on the political purpose of “county administrated by province”, the fiscal capacity of the county is improved due to the successful implementation of the fiscal reform of “county administrated by province”, so that the financial balance sheet at the city level is reshaped to a large extent. Before the implementation of the policy, the municipal government has the right to withhold the fiscal transfer payment of central or provincial governments that should be given to the county government. After the policy is implemented, the channel for the municipal government to transfer some expenditure responsibilities to the county government is cut off. The urban government may be stimulated to further expand land leasing activity to fill this gap under stimulation of a decrease in urban revenue and increase in potential expenditure responsibility caused by the reform of “county administrated by province”. Additionally, the reform of “county administrated by province” not only promotes the county government to reduce the tax burden of the enterprises, but also increases the expenditure on the construction of infrastructure [23].

Thirdly, in terms of financial problem solving, the reform of “county administrated by province” not only increases the revenue of county government and transfer payments between governments, but also expands the responsibility of the county government for expenditure. Some researchers, however, thought that the flattening of government hierarchies has a negative impact on economic performance. The reform of “county administrated by province” not only effectively promotes the fiscal revenue at the county level, but also expands the fiscal expenditure, resulting in difficulty in improving the financial capacity at the county level. The provincial government is difficult to coordinate and monitor the local government due to the expansion of its governance scope, which leads to a decrease in public expenditure and pro-growth expenditure of the county government [24]. From the perspective of the policy of fiscal decentralization, a vicious circle between fiscal increase and economic growth is formed inevitably by increasing fiscal transfer to relieve the fiscal difficulties of the county government or expanding the economic management authorities to improve the economic growth of the county, to further affect the sustainable development of finance and economy of the county.

2.2. Research on the Production Efficiency of Agricultural Enterprises

Few scholars have conducted specific research on TFP of the agriculture-related enterprises. And relevant literature mainly focuses on agriculture TFP. Since the 1970s, the TFP of agriculture-related enterprises has been gradually applied to the estimation and evaluation of agricultural productivity. Hayami [25] analyzed the output elasticity of different input factors and the international differences in agricultural productivity arising therefrom by using the cross-country data from 1957 to 1962 (38 countries including the United States, India, and Japan). Coelli and Rao [26] estimated the TFP of the agriculture-related enterprises of 93 countries from 1980 to 2000 by MPI. Since 1978, the TFP in China has changed greatly. Although the average annual growth rate of TFP in China was 2.11% from 1961 to 2006 [27], the TFP of Chinese agriculture increased at an annual rate of 3% using the DEA method from 1978 to 2008 [28], and the technological changes have become the main reason for the growth of TFP of Chinese agriculture. And the relevant research suggested that inclusive finance, urban expansion, expenditure of agricultural products, and agricultural product insurance significantly contribute to the improvement of TFP [29,30,31,32].

According to literature reviews, most of the literature focused on studying the microeconomic impact of the reform of “county administrated by province”. Some scholars studied the impact of the reform on the economic vitality of the county from the perspective of economic growth. And some discussed the impact of the reform on change in government activities, and some emphasizes on solving the fiscal difficulties at the grassroots level. Few kinds of literature discussed the impact of the reform of “county administrated by province” on TFP of the agriculture-related enterprises from the perspective of enterprises. The TFP of agriculture-related enterprises is an important basis for the measurement of high-quality development of rural industries. Agriculture-related enterprises play a vital role in driving the development of agriculture and rural areas and promoting the modernization of agriculture and rural areas. Consequently, what is the impact of the reform of “county administrated by province” on the production efficiency of micro-enterprises in the county? To answer this question, this paper carried out an empirical study on the impact of the reform of “county administrated by province” on TFP of the agriculture-related enterprises. We used a quasi-natural experiment constructed by the policy of “county administrated by province” and a time-varying difference-differences method to analyze the effect of the policy on the promotion of TFP of the agriculture-related enterprises based on the data of the agriculture-related enterprises in the micro-manufacturing industry. The marginal contributions of this paper include three points. The first is that the reform of “county administrated by province” will inhibit the promotion of the TFP under certain conditions. The second is to test the heterogeneity of the impact of the reform of “county administrated by province” on the TFP. The heterogeneity test found that the reform of “county administrated by province” has a more significantly-negative effect on the agriculture-related enterprises of light industry and in high-grade cities and eastern areas. The third is to explore its impact mechanism and path. The research showed that the urban property prices, labor resource allocation, and banking competition will strengthen the actual inhibition effect of the reform of “county administrated by province” on the TFP of the agriculture-related enterprises.

3. Research Hypothesis

After the reform of “county administrated by province”, the county government obtained fiscal decentralization and power expansion of the economy and may have more capabilities and financial resources to develop the economy of the county and firmly implement the “rural revitalization strategy”, to promote the overall development of urban and rural areas, increase investment in rural areas, improve the rural environment, promote the development of rural economy, and promote the TFP of the agriculture-related enterprises. However, Davoodi and Zou [33] found that fiscal decentralization hurts economic growth through researching the samples including developing and developed countries. They thought that the policy of “county administrated by province” cannot necessarily solve the financial difficulties of the county, and the original system of “county administrated by city” is difficult to adapt to the common economic development of counties and cities, which is because the functional transformation of government lags and cannot be solved simply by adjustment of administrative divisions. The administrative efficiency will be reduced and the burden of the provincial government in the allocation of public goods will be increased due to the excessive span of management resulting from the policy of “county administrated by province”. Based on the analyses mentioned above, this paper puts forward a competitive Hypothesis 1:

Hypothesis 1 (H1a).

The reform of “county administrated by province” is beneficial to the improvement of the TFP.

Hypothesis 1 (H1b).

The reform of “county administrated by province” is not beneficial to the improvement of the TFP.

The distortion of factors allocation at the supply side hurts the TFP. From the supply side, the supply factors include lands, capital, labor, technologies, and intellectual properties. This paper verified the mechanism path of the reform of “county administrated by province” influencing the TFP through the “crowding-out effect”. Due to the limited enterprise resources, a substitute relationship exists between financial investment and entity investment. The enterprises will crowd out the investment in R&D and innovation to a certain extent when carrying out financial investment so as to cause the “crowding-out effect”. Specifically, the enterprises will increase their investments in real estate mainly including commercial and residential lands, and decrease non-land investments after they obtain the credit funds [34] so that the capital will be continuously invested into real estate, and their own development will be reduced certainly in case of limited total capital resource of industrial enterprises, which is not conducive to the development of their main businesses to result in a decrease in capital for R&D and innovation. The expectation brought by high housing prices enlarges the opportunity cost of enterprises’ investment in the real economy and inhibits the innovative behaviors of the enterprises. Many small- and medium-sized enterprises give up industries and invest in real estate due to the huge profits brought by the rapid rise in housing prices. Additionally, the land rent price caused by a rise in housing price changes in the same direction to distort the entrepreneurs’ behaviors and cause more entrepreneurs to join the rent-seeking activities, which is not conducive to enterprise management and innovation investments. From this perspective, the high housing price caused by sustainable high-speed growth of housing prices inhibits the improvement of the TFP. Therefore, this paper proposes the following hypotheses to be tested:

Hypothesis 2 (H2).

The “crowding-out effect” of high housing prices can negatively adjust the impact of CAP reform on the TFP of agriculture-related enterprises.

The literature on resource misallocation suggested that differences in the level of economic development of various countries are largely owing to the difference in TFP [35,36,37]. The rise in TFP is driven by two improvements. The first is the improvement of inside micro productivity brought by technological progress, which has benefited from investment in R&D, technology introduction, deepening of labor division, improvement of management level, and optimal allocation of inside resources. And the second is the improvement of allocation efficiency among enterprises, that is, the production factors flow from the enterprises with low productivity to those with high productivity. If the enterprises with high productivity account for a low share in the market, this indicates that the factors are not effectively allocated to the enterprises, and, thus, there is a low resource allocation among the enterprises. After the reform of “county administrated by province”, the fiscal and economic powers of the government are strengthened, so that such factors as lands, labor, and capital flow to the enterprises with a higher productivity rather than the agriculture-related enterprises, to inhibit the agricultural TFP. Therefore, this paper proposes the following hypotheses to be tested:

Hypothesis 3 (H3).

The allocation of urban resources strengthens the negative adjustment effect of “county administrated by province” on the TFP of agriculture-related enterprises.

For agriculture-related enterprises in China, investment in bank credit plays a major role in financing. The competition in the banking market has an important influence on selecting target enterprises and issuing loans. The banking concentration is employed in the existing literature to measure the competition situation of the banking industry. The banking concentration means that a few banks account for a large market share, which reflects the degree of competition and monopoly in the market. Generally, higher concentration indicates that one or a few banks jointly monopolize the credit market, and the competition in different types of banks is not fierce; otherwise, there is fierce competition. Facing such fierce competition in the banking market, commercial banks have to emphasize selecting enterprises with short cycles, high resource allocation, and high total factor productivity to issue loans. Agriculture-related enterprises are at a disadvantaged position in financing. The banks are cautious to issue loans to them due to slow capital recovery and high inventory holding costs. As a result, a considerable number of agriculture-related enterprises are hard to obtain sufficient external financing aids and face financing constraints. However, the productivity improvement should be supported by financing. Financing difficulties will make the agriculture-related enterprises hard to make optimal operation decisions, optimizing resource allocation efficiency, and, thus, leading to a decline in productivity.

Therefore, this paper proposes the following hypotheses to be tested:

Hypothesis 4 (H4).

In the region where the banking concentration is lower and the competition in commercial banks is more fierce, the policy of “county administrated by province” has a stronger restraint on the TFP of the agriculture-related enterprises, namely, the banking competition negatively adjusts the influence of the reform of “county administrated by province” on the TFP of agriculture-related enterprises.

The mechanism path of the reform of “county administrated by province” influencing the TFP of agriculture-related enterprises is shown in Figure 1.

4. Research Methodology

4.1. Data Sources and Sample Selection

The observation data from 1998 to 2013 are selected in this paper due to their availability (The National Bureau of Statistics only published the data on China’s industrial enterprises from 1998 to 2013, so the data from the same years were remained for matching). The period from 1998 to 2003 is the transitional period for the implementation of the policy of “county administrated by province”. Since 1992, the Chinese government has built the pilot of “increasing power of the county”, and successively realized the reform of “county administrated by province” in 2004. The period from 1998 to 2013, as an intermediate stage of policy implementation, was the stable period that the policy of “county administrated by province” was stably implemented. The power was transited stably from the provincial government to the local government. And there was a balanced relationship between the central and local governments, and the fiscal allocation and economic development were stable. Therefore, the data from 1998 to 2013 can more truly reflect the impact of the implementation of the policy of “county administrated by province” on the TFP. The data of “county administrated by province” are abstracted from the policy documents on the reform of “county administrated by province”, such as the Opinions of the Hebei Provincial People’s Government on expanding the administrative authority of some counties (cities) issued by Hebei Province, and the Notice of Anhui Provincial People’s Government on implementation of fiscal system reform of “county administrated by province”, and a total of 575,229 values were observed at last. Other data used in this paper were from “China Industry Business Performance Data” and Csmar. In data processing, according to Cai and Liu [38], three types of outliers in China Industry Business Performance Data were first eliminated. The first was the samples that violated accounting standards, such as the samples that paid-in capital was less than or equal to 0, the original value of fixed assets was less than 0, the total industrial output value was less than 0, fixed assets were greater than total assets, total assets were less than current assets, and accumulated depreciation was less than depreciation of this year; the second was the samples that an average number of employees was less than 10, and the main business income was less than 5 million; the third was the missing samples such as opening time, administrative code, industry code, and others.

Secondly, we added missing values. In order to retain as many observations as possible, we added missing values for the main key indicators. Brandt et al. [39] sorted out the missing values in the China Industry Business Performance Data in detail. Based on this, we adopted the calculation method of “Industrial added value = total industrial output value-industrial intermediate input + value-added tax” to supplement the missing values of added value. For the condition that the index of industrial output value in 2004 was missed, we used the calculation method of “industrial added value = sales revenue-opening inventory + ending inventory-industrial intermediate input + VAT” to make up this value. We selected “Intermediate input = total output value × main business cost/main business income-total wages payable-depreciation in the current year + finance expenses” to estimate intermediate input.

Then, in order to objectively reflect the value of capital and labor to economic growth, we deflated all nominal prices in the samples based on 1998. Industrial added value was deflated using the price index of fixed asset investment of each province, and the capital stock was deflated using the producer price index. The above two price indices were derived from the China Statistical Yearbook.

Finally, we, based on the relevant industry codes, selected the agriculture-related enterprises from six industries including the agricultural and sideline food processing industry, food manufacturing industry, beverage manufacturing industry, tobacco manufacturing industry, textile industry, and wood processing and wood, bamboo, rattan, brown, and grass manufacturing industry as the research objects. And we focused on the internal mechanism of influence of fiscal decentralization on the total factor productivity of agriculture-related enterprises and determined the effective fiscal decentralization structure through empirical tests.

4.2. Variable Selection and Data Description

(1) Total factor productivity (TFP) of agriculture-related enterprises. The industrial added value of agriculture-related enterprises was used as the index of production capacity, the number of employees of agriculture-related enterprises at the end of the year as the index of labor input, the net fixed assets of agriculture-related enterprises at the end of the year as the index of capital stock, and the intermediate input of agriculture-related enterprises as the proxy variable index. In order to solve the collinearity between the estimation coefficients of the first step of the two TFP calculation methods such as OP and LP, and make necessary amendments to the parameter method, this paper selected the ACF method estimated by the latest semi-parameter and used Stata14. 0 software to estimate the TFP [40]. Among them, according to the accounting standard “net fixed assets = original value of fixed assets-accumulated depreciation”, we took the logarithm of the above indices.

(2) Reform of “CAP (county administrated by province)”. The core explanatory variable of this paper is the variable CAP, which is the interaction term of the dummy variable treat of the treat group and the dummy variable before and after treatment. For the counties and cities that implement the reform of CAP, the variable value is 1 after the implementation of the policy; otherwise, the value is 0. The data on “county administrated by province” are derived from the policy documents on the reform of “county administrated by province” of each province, such as the Opinions of the Hebei Provincial People’s Government on expanding the administrative authority of some counties (cities) issued by Hebei Province, and the Notice of Anhui Provincial People’s Government on implementation of fiscal system reform of “county administrated by province”.

(3) Asset-to-Debt Ratio (ADR). The Asset-to-Debt Ratio of enterprises is an important symbol to measure the level of liabilities and risk degree of enterprises, which will affect the TFP. In order to control the impact of the Asset-to-Debt Ratio on the total factor productivity of agriculture-related enterprises, the asset-liability ratio variable is introduced into the model. The ratio of liabilities to assets is used as a measurement index of asset-liability ratio. The data are derived from the China Industry Business Performance Data.

(4) Take the logarithm of the number of employees at the end of the year (lnlfp). The number of employees at the end of the year has an important impact on the TFP. In order to control the impact of the average number of all employees on the TFP, the logarithmic variable of the average number of all employees is introduced into the model as a measurement index, and the data are from Csmar.

(5) Take logarithm of net fixed assets (lnnfa). Capital input is an important factor affecting the TFP. In order to control the impact of capital input on the TFP, the net fixed assets (NFA) are introduced into the model as a measurement index, and the data are derived from the China Industry Business Performance Data.

(6) Take the logarithm of industrial intermediate inputs (total). In order to control the impact of industrial intermediate input on the TFP, the logarithmic variable of net fixed assets (NFA) is introduced into the model as a measurement index, and the data are derived from the China Industry Business Performance Data.

(7) Take the logarithm of fixed-asset investment (lnnfal). The fixed-asset investment can influence the TFP. In order to control the impact of fixed-asset investment on the TFP, the fixed-asset investment is introduced into the model as a measurement index, and the data are derived from the China Industry Business Performance Data.

(8) Gross output value per capita (lnpgdp). The regional economic development level has an important impact on the TFP. In order to control the impact of the regional economic development level on the TFP, the variable of the regional economic development level is introduced into the model as a measurement index, and the data are from Csmar.

(9) Urban Property Prices (upp). The logarithm of urban property prices (upp) is introduced as the adjusting variable, and the data are from the statistical database of the China Economic Network.

(10) Efficiency of resource allocation among enterprises in different cities ($Y_c$). The efficiency of resource allocation among enterprises in different cities is introduced as the adjusting variable, and the data are from the China Industry Business Performance Data.

(11) Banking concentration (HHI). In order to verify the effect of bank competition on the impact of “county administrated by province” on the TFP, the concentration degree of the banking industry is introduced as a moderator variable, and the data are from the China Banking and Insurance Regulatory Commission.

The definitions of each variable are shown in

Table 1. Variable names and calculation caliber.

Variable Type	Symbol	Variable Meaning	Variable Declaration
Dependent variable	TFP	Total factor productivity of agriculture-related enterprises	ACF method
Independent variable	CAP	Reform of “county administrated by province”	From the province about the “county administrated by province” A policy document for the reform
Control variables	ADR	asset-liability ratio	Debt/assets
	Lnlfp	Number of employees at the end of the year	Logarithmic value
	Lnnfa	net fixed assets	Logarithmic value
	Lngii	Industrial intermediate input (total)	Logarithmic value
	Lnfal	fixed-asset investment	Logarithmic value
	Lnpgdp	Gross output value per capita	Logarithmic value
Moderator variables	upp	Urban Property Prices	Logarithmic value
	Yc	The efficiency of resource allocation among enterprises in various cities	Labor force among urban enterprises Resource allocation efficiency
	HHI	Banking concentration	Bank branch Herfindahl index

.

4.3. Model Setting

In order to verify the impact of the CPA reform on the TFP, this paper, based on the experience of Kudamatsu [41], used the asymptotic time-varying difference-differences method and two-way fixed Effects (TWFE) method to estimate the TFP, and added the time and individual fixed effects, and then clustered to enterprises. Considering that the CPA reform is implemented in different periods in different regions, in order to investigate the impact of the reform on the TFP, this paper took reference with Hoynest et al.’s [42] transformation idea of “natural experiment”, did not set a unified dummy variable Treat and a dummy variable before and after treating Post, and only contained the interaction term CAP into the model, and set following baseline regression model:

$$TF{P}_{it}={\alpha }_0+{\alpha }_1{CAP}_{it}+{\alpha }_i{X}_{it}+{\gamma }_t+u_i+{\epsilon }_{it}$$

(1)

where ${\alpha }_0$ is the model constant term, $TF{P}_{it}$ is the explained variable: total factor productivity of agriculture-related enterprises, subscripts i and t represent the enterprise and the year; ${CAP}_{it}$ is the explanatory variable: county administrated by province, $X_{it}$ is a series of control variables at the level of enterprises and prefecture-level cities, including asset-liability ratio (ADR), the logarithm of the number of employees at the end of the year (lnlfp), the logarithm of net fixed assets (lnnfa), the logarithm of total industrial intermediate input (lngii), and the logarithm of fixed asset investment (lnfal); ${\gamma }_t$ is the time-fixed effect; $u_i$ is the individual (enterprise) fixed effect; ${\epsilon }_{it}$ is the random disturbance term.

5. Results and Discussion

5.1. Descriptive Statistics and Benchmark Regression

Table 2. Variables description.

Variable	N	Mean	SD	p50	Min	Max
TFP	530923	0.784	1.067	0.796	−8.378	7.399
CAP	575229	0.055	0.227	0	0	1
ADR	575215	0.570	0.433	0.555	−1.851	104.4
Lnlfp	574967	4.868	1.105	4.868	0	11.93
Lnnfa	574137	8.619	1.784	8.631	0	17.67
Lngii	542806	10.22	1.439	10.17	0	19.07
Lnfal	303967	8.247	1.975	8.240	0	18.32
Lnpgdp	575229	1.299	0.628	1.227	0.152	3.895

shows the descriptive statistical results of the main variables, from which no outliers occur. The core explained variable is total factor productivity (TFP) of agriculture-related enterprises.

The impact of the CAP reform on the TFP of agriculture-related enterprises by the panel data fixed effect model was estimated in accordance with the measurement model (1) designed in this paper. The control variables at the enterprise and city level were added to column (1), and the individual and time-fixed effects were carried out in columns (2) to (4), and the standard errors were clustered at the enterprise level. The estimation results of the benchmark model for the CAP reform are shown in

Table 3. Baseline Estimation Tests.

	Fixed Effects of Panel Data
	(1)	(2)	(3)	(4)
	TFP	TFP	TFP	TFP
PMC	−0.098 ***	−0.071 ***	−0.044 ***	−0.044 ***
	(−14.10)	(−9.90)	(−4.00)	(−2.94)
ADR	−0.265 ***	−0.263 ***	−0.162 ***	−0.162 ***
	(−55.73)	(−55.05)	(−22.70)	(−16.03)
Lnlfp	−0.324 ***	−0.309 ***	−0.345 ***	−0.345 ***
	(−171.10)	(−158.23)	(−119.95)	(−82.20)
Lnnfa	−0.171 ***	−0.169 ***	−0.164 ***	−0.164 ***
	(−102.47)	(−101.99)	(−65.01)	(−47.13)
Lngii	0.602 ***	0.604 ***	0.490 ***	0.490 ***
	(375.83)	(363.71)	(200.32)	(75.72)
Lnfal	0.051 ***	0.052 ***	0.030 ***	0.030 ***
	(42.07)	(43.43)	(20.49)	(17.74)
Lnpgdp	−0.012 ***	−0.005	0.027 *	0.027
	(−4.21)	(−1.32)	(1.80)	(1.31)
Enterprise fixed effect	no	yes	yes	yes
Year fixed effect	no	no	yes	yes
Clustering enterprises	no	no	no	yes
Constant	−2.472 ***	−2.713 ***	−1.315 ***	−1.315 ***
	(−182.29)	(−176.48)	(−44.83)	(−24.48)
Observations	292,423	292,423	292,423	292,423
Within R-squared			−0.135	0.246

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. ***, ** and * represent significance at the levels of 1%, 5% and 10% respectively.

. From the estimation results, we found that the estimated coefficient of impact of CAP reform on the TFP was significantly negative at the level of 1% when various control variables were added and individuals and time were fixed, that is, the CAP reform did significantly reduce the TFP, and, thus, the hypothesis H1a was verified.

The asymptotic time-varying difference-differences method was used in this paper. To verify the robustness of this effect, its parallel trend test and placebo test should be conducted respectively.

5.2. Robustness Test

5.2.1. Placebo Test: Randomly Select the Test Group

Ideally, when our policy is not affected by unobservable factors, that is, when the policy is completely exogenous, a consistent estimator of the coefficient can be obtained using OLS estimation. However, in real life, our policy will be affected by various factors, and we cannot exhaust all the control variables, so the results therefrom may be biased.

In order to further demonstrate that the slowdown in the promotion of the TFP is mainly caused by the implementation of the CAP reform rather than other unobservable factors. This paper used the analytical thinking of Rachel et al. [43] to conduct a placebo test; that is, in all the county samples within the research range, the counties were randomly divided into a test group and control group, and the number of counties should be the same with the counties newly established as “counties administrated by province” each year, and then the random logarithm model was used to repeatedly estimate model (1). A total of 1000 samples were randomly and repeatedly taken to obtain 1000 estimation coefficients of the core explanatory variables ${CAP}_{}$, with the mean distribution of 0.0013. And its normal distribution kernel density is shown in Figure 2.

The coefficient estimated in column 4 of Table 3 was −0.044 (outside Figure 2). According to Figure 2, the estimated value (−0.044) is significantly different from the coefficient estimated in the placebo test, thus confirming that the inhibition effect of CAP reform on the TFP improvement is not derived from unobvious factors.

5.2.2. Multidimensional Fixed Effects Test

In order to further strictly control the false correlation between the implementation of “CAP” reform and the TFP of agriculture-related enterprises caused by missing variables, this paper adopted the multi-dimensional fixed effect model to perform robustness tests based on existing individual fixed effects and time-fixed effects. The interaction terms of industry-fixed effect and time-fixed effect, as well as the interaction terms of industry-fixed effect and county-fixed effect, were further added into the model to estimate the impact of the implementation of the CAP reform on the TFP. The regression results were shown in

Table 4. Multidimensional fixed effects of panel data.

	(1)	(2)
	TFP	TFP
CAP	−0.037 **	−0.044 ***
	(−2.50)	(−2.86)
Control	yes	yes
Year fixed effect	yes	yes
Enterprise fixed effect	yes	yes
Clustering enterprises	yes	yes
Industry * Year fixed effect	yes	no
Industry * region fixed effect	no	yes
Constant	−1.300 ***	−1.171 ***
	(−20.73)	(−18.47)
Observations	257,594	257,320
Within R-squared	0.699	0.688

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. ***, ** and * represent significance at the levels of 1%, 5% and 10% respectively.

. According to empirical analysis, the regression results added with industry- and region-fixed effects were generally consistent with the benchmark regression model; that is, the impact of the CAP reform on the TFP was significantly negative at a 1% level. However, according to empirical analysis, the regression results added with industry and region fixed effects were generally consistent with the benchmark regression model; that is, the impact of the CAP reform on the TFP was significantly negative, at a 5% level, and the coefficient was improved.

5.2.3. Replacement Cluster Robust Standard Error

There is often an autocorrelation among the disturbance terms of the same enterprise in different periods when replacing the cluster robust standard error panel data. In this case, it is problematic to use ordinary standard error or heteroscedasticity robust standard error, and the statistic t (coefficient value/standard error) will also fail. In this paper, the cluster robust standard error was applied to the panel data. All the observed values in different periods of each enterprise constituted a “cluster”. The observed values in the same cluster were correlated with each other, and the observed values between different clusters were not correlated. From columns (1) to (3), we selected more stringent clusters to the city, county, and industry levels, and the regression results were shown in

Table 5. Clustering of cities or counties.

	(1)	(2)	(3)
	TFP	TFP	TFP
CAP	−0.082 **	−0.082 ***	−0.080 ***
	(−2.24)	(−2.62)	(−3.35)
control	yes	yes	yes
Year fixed effect	yes	yes	yes
Enterprise fixed effect	yes	yes	yes
Clustering enterprises	yes	no	no
Clustering county	no	yes	no
Clustering industry	no	no	yes
Observations	292,423	292,423	286,760
Within R-squared	0.508	0.508	0.507

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. *** and ** represent significance at the levels of 1% and 5%, respectively.

. According to this table, after clustering cities, counties, and industries, the estimated coefficients of the CAP reform on the TFP were significantly negative at 5% and 1% respectively; that is, the CAP reform significantly inhibited the TFP of agriculture-related enterprises.

5.3. Heterogeneity Analysis

5.3.1. Heterogeneity Analysis at the City Level

In different regions and cities with different levels of economic development, there is a difference in the ideas and internal drivers of the CAP reform for the development of agriculture-related enterprises in the region. The industry development design and investment attraction adopted by the “county administrated by province” where the cities with a stronger economic foundation and higher administrative level are located may be more inclined to the enterprises which are more profitable in the faster way due to the expensive land price and the impact of the political and economic environment, while the agriculture-related enterprises are located in inferior position due to less and slow profits creation, and, thus, the positive impact of the TFP is suppressed.

Therefore, this paper distinguished between cities with higher administrative levels (relatively higher levels of economic development) and general prefecture-level cities and investigated the impact of the CAP reform on the TFP. Specifically, 4 municipalities directly under the Central Government, 26 provincial capitals, and 5 cities specifically designated in the state plan were taken as the samples of cities at higher administrative levels, and the other cities were the samples of cities at lower administrative levels, to test the city-level heterogeneity of impact of the CAP reform on the TFP of agriculture-related enterprises.

According to columns (3) and (4) of

Table 6. Heterogeneity analysis at the city level.

Variables	(1)	(2)
	Cities
	High-Level	Low-Level
CAP	−0.180 ***	−0.025
	(−3.17)	(−1.63)
control	yes	yes
Year fixed effect	yes	yes
Enterprise fixed effect	yes	yes
Clustering enterprises	yes	yes
Constant	−0.847 ***	−1.456 ***
	(−5.98)	(−24.99)
Observations	59,412	233,011
Within R-squared	0.183	0.267

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. *** and ** represent significance at the levels of 1% and 5%, respectively.

, the CAP reform significantly inhibited the TFP of high-level cities but had no significant impact on cities at low administrative levels. As a result, the development of agriculture-related enterprises had no advantages in high-grade cities.

5.3.2. Regional Heterogeneity Analysis

China has a vast territory, so different regions have great differences in factor endowment, economic base, geographical location, and historical conditions, as well as the agricultural technology, capital accumulation, and supporting degree of agriculture-related industries, thus leading to a regional heterogeneity in the degree of the TFP caused by the CAP reform. Therefore, we divided the samples into eastern, central, and western regions in terms of geographical locations, and tested the regional heterogeneity of the CAP reform on the TFP of agriculture-related enterprises according to Equation (1). The results were shown in

Table 7. Regression results of regional heterogeneity analysis.

	West	Central	East
	TFP	TFP	TFP
CAP	0.043	−0.014	−0.053 ***
	(1.09)	(−0.48)	(−2.82)
control	yes	yes	yes
Year fixed effect	yes	yes	yes
Enterprise fixed effect	yes	yes	yes
Clustering enterprises	yes	yes	yes
Constant	−1.058 ***	−0.834 ***	−1.519 ***
	(−5.35)	(−6.43)	(−25.07)
Observations	24,363	58,997	209,063
Within R-squared	0.247	0.205	0.266

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. *** and ** represent significance at the levels of 1% and 5%, respectively.

.

According to the estimated results in

Table 8. Results of the interaction between CAP and Urban Property Prices.

	(1)	(2)	(3)	(4)
	TFP	TFP	TFP	TFP
CAP * upp	−0.192 ***	−0.153 ***	−0.154 ***	−0.154 ***
	(−10.31)	(−8.21)	(−6.92)	(−5.19)
control	yes	yes	yes	yes
Year fixed effect	no	yes	yes	yes
Enterprise fixed effect	no	no	yes	yes
Clustering enterprises	no	no	no	yes
Constant	−2.409 ***	−2.613 ***	−1.008 ***	−1.008 ***
	(−143.10)	(−145.21)	(−24.83)	(−15.75)
Observations	216,281	216,281	216,281	216,281
Within R-squared			−0.209	0.233

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. ***, ** and * represent significance at the levels of 1%, 5% and 10% respectively.

, in terms of the division of geographical locations, the impact of the CAP reform in the eastern regions was significantly negative on the TFP at a 1% level, while not significant in the central and western regions. Specifically, with an increase of 1 unit in the CAP reform, the TFP in the eastern regions will be reduced by 0.053 units.

5.4. Mechanism Test

5.4.1. The Crowding Out Effect Caused by Housing Prices Rising

The rise in housing prices may increase the confidence of agriculture-related enterprises in financial investments and induce them to invest in the real estate market to crowd out their investments in entities, thereby affecting their investments in technological innovation. The crowding-out effect caused by housing prices may profoundly affect the implementation effect of the CAP policy. To this end, we constructed an interactive effect model for verification:

$${TFP}_{it}={\alpha }_0+{\alpha }_1{CAP}_{it}\cdot {upp}_{it}+{\alpha }_i{X}_{it}+{\gamma }_t+{\mu }_i+{\epsilon }_{it}$$

(2)

where ${upp}_{it}$ is the logarithm of city housing price, which is derived from the statistical database of China’s Economic network, and the definitions of other variables are the same as those of the benchmark regression model. Table 7 reflected the regression estimation results of the interaction terms of CAP and urban property prices. According to this table, when such variables as time, individual fixed effect, and standard error cluster are added to the enterprise level, the TFP is significantly negative. The result suggested that, compared with areas with lower housing prices, there was a more significant inhibition of CAP on TFP of agriculture-related enterprises in areas with higher housing prices, that is, the crowding out effect had a negative regulating effect on the influence of CAP on TFP of agriculture-related enterprises, indicating that the rise in housing prices may lead the management to make a short-sighted decision to invest in real estate, to crowd out the technological innovation investment and reduces the TFP of agriculture-related enterprises.

5.4.2. Resource Mismatch Effect

Olley and Pakes [44] proposed a method that evaluates inter-enterprise resource allocation efficiency so that the resource allocation efficiency index calculated by this method can better measure the resource mismatch at the regional level [45]. Given this, we decomposed the productivity of industry j in city c weighted by enterprise factor shares as follows:

$${\Omega }_{cj}={\sum }_i{\theta }_{cji}{\omega }_{cji}=\overline{w_{cj}}+{\sum }_i\left({\theta }_{cji}-\overline{{\theta }_{cj}}\right)\left({\omega }_{cji}-\overline{w_{cj}}\right)$$

(3)

where ${\omega }_{cji}$ is the productivity of enterprise I, ${\theta }_{cji}$ is the factor share of enterprise i of industry j in the city c, $\overline{w_{cj}}$ is the simple average productivity of all enterprises of industry j in the city c, $\overline{{\theta }_{cj}}$ is the simple average factor share of all enterprises of industry j in the city c. In this paper, the city refers to the entire prefecture-level administrative unit, while the industry is divided based on the two-digit industry classification. The first term in Formula (3) reflects the level of enterprise micro-productivity, and the second term is the covariance between enterprise factor share and productivity (OP covariance).

Bartelsman et al. [46] found that, in various industries of a city, the resource allocation efficiency among enterprises reflects the share of the number of employees of enterprises in the total number of employees of industry j in city c. A higher resource allocation efficiency index indicates that a larger proportion of resources is obtained by the enterprise with high productivity, which implies that an efficient resource allocation is implemented. Using the data of China Industry Business Performance Data from 2003 to 2007, we can calculate the resource allocation efficiency index at the city-industry level discussed in this study. Specifically, we can calculate the share of the number of employees of each enterprise I of industry j in city c in the total number of employees, and then calculate the resource allocation efficiency index $Y_{cj}$ at the city-industry level.

$$Y_{cj}=\sum \left({\theta }_{cji}-\overline{{\theta }_{rj}}\right)\left({\omega }_{cji}-\overline{{\omega }_{cj}}\right)$$

(4)

Furthermore, we took the share of the labor force of each industry in a city as weight and summed the resource allocation efficiency index $Y_{cj}$ at the city-industry level to the city level to obtain the index which is used to measure the resource allocation efficiency index among enterprises in various cities $Y_c$:

$$Y_c={\sum }_j{\theta }_{cj}{Y}_{cj}$$

(5)

Table 9. Results of the interaction between CAP and resource allocation efficiency.

	(1)	(2)	(3)	(4)
	TFP	TFP	TFP	TFP
CAP * Yc	−0.055 ***	−0.055 ***	−0.042 ***	−0.042 ***
	(−9.24)	(−9.25)	(−5.30)	(−3.49)
control	yes	yes	yes	yes
Year fixed effect	no	yes	yes	yes
Individual fixation effect	no	no	yes	yes
Clustering enterprises	no	no	no	yes
Constant	−2.478 ***	−2.697 ***	−1.317 ***	−1.317 ***
	(−177.66)	(−174.14)	(−44.12)	(−24.25)
Observations	291,691	291,691	291,691	291,691
Within R-squared			−0.136	0.246

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. ***, ** and * represent significance at the levels of 1%, 5% and 10% respectively.

showed the regression estimation results of CAP and enterprise resource allocation efficiency. The proportion of the number of employees of industry j in city c to the total number of employees in all industrial enterprises in city c was taken as a weight. After clustering the time, individual fixed effect, and standard error to the enterprise level, it is found that the TFP had a significant negative effect, indicating that the allocation of urban resources strengthens the negative regulation effect of CAP on the TFP of agriculture-related enterprises. Specifically, compared with non-agricultural enterprises with higher efficiency of urban resource allocation, the agriculture-related enterprises did not have an advantage in resource regulation caused by the CAP reform, so their total factor productivity was lower.

5.4.3. Influence of Banking Concentration

Two kinds of methods were used in this study to evaluate the competition in the banking industry. The first was the “nonstructural method”, which mainly adopted the new empirical industrial organization method based on the Panzar–Rosse model (H-statistical method) [47,48,49,50]. The second was the “structural method”, where the Herfindahl index of banking concentration was a representative index widely used for concentration measurement; for example, Petersen and Rajan [51] used HHI as a measure of competition in their empirical research, and Boot and Thakor [52] used the number of banks and the HHI therefrom to describe the degree of competition in their model. Concerning the research of Degryse and Ongena [53] and Chong et al. [54], this paper adopted the number of bank branches to calculate the Herfindahl index. As the number of commercial bank branches can better reflect the competition level of banks, the number of commercial bank branches is more accurate than the competition index of the financial industry and the Lerner index, in case the loan amounts issued by the commercial bank branches are not available. It is noted that most branches and head offices of commercial banks can issue loans through their business department, so we calculated the Herfindahl index of the head office, branches at all levels, sub-branches, business offices, branch offices, and savings offices. Through collating the original data of the China Banking and Insurance Regulatory Commission, information on 168, 841 branches of commercial banks holding financial licenses in China were obtained. Using manual and Stata matching, bank branches were matched to 1,968 counties. For bank branches that cannot be matched, we checked their financial license number on their official website and then re-matched the data. Finally, we obtained the banking concentration index of prefecture-level cities by summing up the data of each county. The formula of the urban banking concentration was shown below:

$$HH{I}_{it}={\sum }_{\alpha =1}^{N_{\mathrm{m}}}{\left({\mathrm{branch}}_{\alpha m}/{\sum }_{\alpha =1}^{N_{\mathrm{m}}}{\mathrm{branch}}_{\alpha m}\right)}^2$$

(6)

In Formula (2), ${\mathrm{branch}}_{\alpha m}$ is the $\alpha$ number of branches of alpha bank in city m, $N_{\mathrm{m}}$ is the number of all types of commercial banks in city m. The range $HH{I}_{it}$ is from 0 to 1. The closer the value is to 0, the lower the concentration and the fiercer the competition among banks. If the value is equal to 1, indicating that the credit market is monopolized by a single bank. In the samples, the mean value of the concentration was 0.19, with the maximum value of 0.71, and the minimum value of 0.06.

Table 10. Results of the interaction between CAP and Banking concentration.

	(1)	(2)	(3)	(4)
	TFP	TFP	TFP	TFP
CAP * HHI	0.532 ***	0.409 ***	0.482 ***	0.482 ***
	(6.15)	(4.76)	(3.87)	(2.69)
control	yes	yes	yes	yes
Year fixed effect	no	yes	yes	yes
Individual fixation effect	no	no	yes	yes
Clustering enterprises	no	no	no	yes
Constant	−2.500 ***	−2.798 ***	−1.297 ***	−1.297 ***
	(−161.51)	(−162.28)	(−40.58)	(−22.82)
Observations	291,691	291,691	291,691	291,691
Within R-squared			−0.136	0.246

Note: t-values are reported in parentheses. The robust standard errors are clustered at the enterprise level. ***, ** and * represent significance at the levels of 1%, 5% and 10% respectively.

reflected the regression estimation results of the CAP and the concentration. According to this table, when such variables as time, individual fixed effect, and standard error cluster were added to the enterprise level, the TFP was significantly positive.

As a result, the concentration weakens the inhibition effect of the CAP reform on the TFP, or the banking competition can promote the influence of the CAP reform on TFP, that is, the H4 is practical.

6. Conclusions

This paper empirically tested the influence of CAP reform on the TFP of agriculture-related enterprises using the time-varying difference-differences method and two-way fixed effects (TWFE) method, based on the data of agriculture-related enterprises and of CAP in China from 1998 to 2013. With the inclusion of various control variables and the imposition of fixed effects for individuals and time, our regression revealed the CAP reform can significantly inhibit the TFP of agriculture-related enterprises at a 1% level of significance. In our analysis of heterogeneity, we have observed that the implementation of the CAP reform had a significantly negative impact on the total factor productivity (TFP) within the light industries of the manufacturing sector at a 1% level of significance. However, this impact was not found to be statistically significant for agriculture-related enterprises in heavy industries. Furthermore, the CAP reform was found to significantly hinder the TFP of high-level cities, but it did not have a significant impact on cities at a lower administrative level. Additionally, when considering regional differences, the CAP reform was found to have a significantly negative impact on the TFP in the eastern regions, whereas no significant impact was observed in the central and western regions.

In addition, according to the mechanism analysis, the “crowding out effect” negatively regulates the influence of the CAP reform on the TFP of agriculture-related enterprises, and the rapid growth of urban property prices inhibits the improvement of the TFP of agriculture-related enterprises. The allocation of urban resources strengthens the negative regulating effect of the CAP reform on the TFP of agriculture-related enterprises. The inhibition effect of the CAP reform on the TFP of agriculture-related enterprises is regulated by the banking competition, and the more intense the competition, the stronger the inhibition effect.

After a series of robustness tests, the main conclusions mentioned above are still practical.

7. Policy Implication

According to the empirical result, the inhibition effect of CAP policy on the TFP of agriculture-related enterprises will be strengthened by the urban property prices, labor resource allocation, and banking competition.

Firstly, the financial and economic power of county-level governments under the CAP policy has been increased, but under the pressure of local competition, local governments often have strong incentives to interfere with the land and financial markets, thus affecting the allocation of land, labor, and other resources among enterprises. And facilitation of financing and the resources (land, labor) allocation is more conducive to those non-agriculture-related enterprises to create high economic benefits in the short term. Secondly, the “CAP” policy aggravates the distortion of resources, and the low production efficiency caused by the imbalance of resource allocation is not conducive to the long-term development of agriculture-related enterprises. As a result, the government should correctly direct the resource allocation, and prevent the risks caused by the misallocation of land, labor, capital, and other resources while giving full play to the resource allocation of the market mechanism. Thirdly, agriculture-related enterprises are vulnerable in the financing market, and, thus, the government should implement the rural revitalization strategy and increase support for the transformation and upgrading of agriculture-related enterprises by providing financing support, tax relief, and other means. Fourthly, the government should encourage agriculture-related enterprises to perform technological research and development and stimulate them to innovate their technologies through financial subsidies, tax incentives, or other measures.

Some deficiencies still exist in this paper. First of all, considering the data availability and the actual situation of the CAP reform, this research is limited spatially to a certain extent because it mainly studies the influence of CAP reform on agriculture-related enterprises based on the data from 1998 to 2003. And if we can obtain the county-level financial data at the national level after 2013 and explore the long-term effects of the CAP reform on agriculture-related enterprises, then conclusions obtained therefrom will be more referential for the universality of the conclusions and the perfection of local financial management systems. At present, the CAP policy is still in the pilot stage, and the government still has bright prospects for deepening the reform of the financial systems, implementing the decentralization of powers, and improving the administrative and financial systems. Agriculture-related enterprises are the key force to promote rural revitalization, and the promotion of the development of the TFP of agricultural enterprises needs to be explored deeply, which may become a subject to be further studied. Finally, the findings of this paper are derived from the reform of the experience of China in a specific period, and their reliability needs to be further tested by relevant studies from other perspectives.