Analysis of ABC Model for Optimizing Financial Structure of Universities in Government Accounting Comprehensive Budget Performance Management

The implementation of a comprehensive budget performance management system is a major change in the management of the national budget, and an important initiative for the establishment of a modern financial system and the improvement of national governance capacity. Government accounting is the foundation of modern financial system and the basic guarantee of national financial governance [1-4]. Budget performance management helps to promote government governance, deepen the reform of government accounting, and facilitate the effective management of budget execution [5-7]. This process requires an in-depth analysis of the relationship between the government accounting system and the basis of accounting before complementarities can be exploited and joint efforts can be made to achieve co-development, leading to more modernized public governance [8-11]. In government accounting comprehensive budget performance management, it is of great significance to use ABC model to analyze the optimization of financial structure of colleges and universities.

With the continuous advancement of China’s education system reform, the catalytic effect of market-oriented factors on the reform of colleges and universities has become more and more obvious, and colleges and universities have gradually become legal entities running schools independently for the society [12-14], with the diversified development of the internal economy and the increasingly complex external economic relations, and the funding for running schools is no longer based on a single financial allocation, and the diversified financing channels have gradually taken shape [15-18]. Changes in the internal and external financial management environment require universities to further standardize their financial activities and improve their financial management level, and at the same time, the involvement of all parties in society in university governance is an inevitable trend in the development of the university, focusing on the needs of different stakeholder groups including government representatives, students and parents, alumni, scientific research institutions, donors, etc. [19-22]. In this context, a comprehensive, systematic and in-depth analysis of the financial structure of universities, so as to achieve the optimization of the financial structure of universities, can help universities to better understand their own financial situation, and provide a basis for future development planning [23-26].

In this paper, a model based on the breakpoint regression analysis method is used to investigate the impact of the ABC model optimization of university financial structure on government budget performance management. The hypothesis of the problem was formulated, and after selecting the relevant variables, the variables were tested for covariance for subsequent analysis. OLS regression method was used to initially obtain the impact of ABC model optimization in the financial structure of universities on local administrative performance. Then the breakpoint regression analysis model is applied to further deepen the influence mechanism between the two variables. The robustness of the assumptions and the specific effect of this paper are verified by robustness analysis and heterogeneity analysis.

2

Modeling the impact of financial structure optimization on budget performance

2.1

ABC model

According to the ABC model [27], attitude consists of three parts: affective, behavioral and cognitive, but the three have different positions and functions in the attitude system.

Among them, cognition is an objective evaluation of the attitude object by the attitude subject, which is a kind of cognition of good or bad. Emotion is based on cognition, the degree of the attitude subject’s preference for the attitude object, such as like or hate, is a kind of feeling. Behavior is generally expressed in terms of behavioral tendencies, and attitudinal behavior is a preparatory state for actual behavior, which does not necessarily result in actual behavior, but only indicates an intention to take some kind of action.

Among the parts that make up the attitude, cognition is the foundation, whether there is a bias in cognition will directly affect the directionality of the attitude, so be sure to maintain a correct cognition of things. Emotion is the key, the emotional preference for the attitude object, will directly affect the subject’s judgment of the object, emotion is the driving force of the composition of the attitude. Behavior indicates the directionality of the attitude, has a guiding role, and affects the subject’s behavioral response. The three are closely linked, interact and harmonize to form a complete organism. The higher the degree of coordination among the three elements, the more stable the attitude is. Attitude is an internal psychological experience of people, so it can not be directly observed, and can only be judged through people’s behavioral performance (including language, expression, action and other manifestations).

In this paper, with the help of ABC model, we complete the optimization of the financial structure of colleges and universities, find out the weak points, and improve the level of comprehensive budget performance management of government accounting through the efforts of each participant.

2.2

Breakpoint regression design method

2.2.1

Applicability of breakpoint regression methods

The concept of regression design with breakpoints (RDD) [28] has been refined and developed to attract the attention of many experts and scholars, with the most common applications being the evaluation of optimization strategies in the fields of education, economics or social sciences. Because breakpoint regression treats the sample as a randomized experiment, it is possible to identify the implementation effect of the optimization strategy in this way. Breakpoint regression is designed to use a certain driver as the individual’s assigned variable, in which the initiation of the optimization strategy is the breakpoint, and a dummy variable is introduced into the model to represent whether the optimization strategy has been implemented or not, and on the left side of the breakpoint the dummy variable is usually assigned a value of 0, in which case the individual does not receive the treatment effect of the optimization strategy. In the left side of the breakpoint, the dummy variable is generally assigned a value of 0, when the individual does not receive the treatment effect of the optimization strategy. Conversely the dummy variable is assigned a value of 1, when the individual is affected by the optimization strategy. If the optimization strategy is regarded as an influencing factor that changes suddenly (the optimization of the ABC model of the financial structure of higher education institutions meets this condition), the difference between the optimization strategy and some other continuously changing confounding variables (both observable and unobservable) can be identified according to this method, so that the impact on the object of study can be analyzed.

The first characteristic of breakpoint regression is the presence of “jumps” at breakpoints, which show discontinuities in the outcome variable, and the direction and magnitude of such jumps are a direct measure of the causal effect of the results of the individual treatments in the vicinity of the breakpoints. The second characteristic of breakpoint regression is local randomness, which is assumed to be based on the premise that individuals falling into either the treatment or control group near the breakpoint are completely random, and that individuals receiving the treatment and those not receiving the treatment are identical on all dimensions of the variable attributes, so that any subsequent differences between individuals can be assumed to be caused by acceptance of the optimization strategy.

Breakpoint regression can be categorized into exact breakpoint regression (SRD) and fuzzy breakpoint regression (FRD) based on the difference in the probability of an individual receiving treatment at a breakpoint. Exact breakpoint regression occurs if every individual larger than the threshold receives treatment and no individual smaller than the threshold receives treatment, i.e., the probability that an individual receives treatment at the breakpoint jumps from 0 to 1. If the probability of an individual getting treated at the breakpoint jumps from a to b, where 0<a<b<1, then it is FRD. The research object of this paper is the level of comprehensive budget performance management of government accounting in China, and the optimization of ABC model of financial structure of colleges and universities starts to be practiced at a certain moment, and the individual is driven by the variable of time allocation, and the probability of the individual to start practicing the optimization strategy at the breakpoint will jump from 0 to 1. Therefore, we take the exact breakpoint regression (SRD) as the theoretical basis of this paper.

2.2.2

Model construction based on breakpoint regression analysis

In this paper, we mainly use the RDD model to assess the impact of the optimization of the ABC model of university financial structure on the level of comprehensive budget performance management of Chinese government accounting, based on the theoretical foundation of the previous paper, we take the date when the optimization starts to be implemented as the breakpoint, which is set to c. We introduce the dummy variable D to indicate the optimization strategy, the control group on the left side of the breakpoint, which is the observed value of the individuals before the optimization starts, and set to 0, and the treatment group on the right side of the breakpoint, which is the observed value of the individuals after the optimization starts and set to 1. The right side of the breakpoint is the treatment group, which is the individual observation value after the optimization of the ABC model of college financial structure starts, and is set to 1. Assuming that the allocation variable is t, and the allocation variable is the basis for judging whether an individual enters the treatment group or not, in this paper, the individuals are divided into the control group and the treatment group according to the time, and t is the time. The treatment effect is the effect of the optimization of the ABC model of university financial structure on the level of government accounting comprehensive budget performance management, and if the regression results show the existence of mutation effect at the breakpoint, it proves that the optimization strategy will have an effect on the dependent variable.

According to the basic principle of quasi-natural experiment, when the individual is affected by the only influence in the external environment, the sample individuals will be randomly divided into two groups at t = c before the experiment (Y₀) and after the experiment (Y₁), at this time, the treatment effect is expressed using the changes produced by the individual before and after the optimization is initiated (Y₀ − Y₁), and the average treatment effect (ATE) can be used as a general estimation of the causal effect of the optimization. 1 $A T E = E (Y_{1} | t = c) - E (Y_{0} | t = c) = \lim_{t \to c^{+}} E (Y | t = c) - \lim_{t \to c^{-}} E (Y | t = c)$ $$ATE = E({Y_1}|t = c) - E({Y_0}|t = c) = \mathop {\lim }\limits_{t \to {c^ + }} E(Y|t = c) - \mathop {\lim }\limits_{t \to {c^ - }} E(Y|t = c)$$

However, this paper cannot exclude the possibility that there are still other optimization strategies or mutation factors in the same period of the ABC model optimization of college financial structure, and if the problem of omitted variables occurs, the results of the optimization strategy intervention treatment effect will not be credible. So the next narrowing of the sample interval, the average treatment effect is limited to the threshold range, find a suitable bandwidth h near the breakpoint t = c, it can be assumed that in the threshold range of (c − h, c + h), the sample is not affected by confounding factors in addition to the intervention of the ABC model optimization of the financial structure of colleges and universities, at this time, you can replace the ATE with the local average treatment effect (LATE): 2 $L A T E = \lim_{t \to c^{+}} E (Y | t = c) - \lim_{t \to c^{-}} E (Y | t = c), c - h \leq t \leq c + h$ $$LATE = \mathop {\lim }\limits_{t \to {c^ + }} E(Y|t = c) - \mathop {\lim }\limits_{t \to {c^ - }} E(Y|t = c)\:,\:c - h \le t \le c + h$$

In this paper, individuals all accept the action of the optimization strategy after the beginning of the optimization strategy, and the probability of entering the treatment group jumps from 0 to 1. Based on the theory of exact breakpoint regression, OLS estimation [29] can be performed on both sides of the breakpoint c respectively for comparison: 3 $Y_{i} = α_{L} + β_{L} (t_{i} - c) + δ_{L} \cdot X + ε_{i, L}$ $${Y_i} = {\alpha_L} + {\beta_L}({t_i} - c) + {\delta_L} \cdot X + {\varepsilon_{i,L}}$$ 4 $Y_{i} = α_{R} + β_{R} (t_{i} - c) + δ_{R} \cdot X + ε_{i, R}$ $${Y_i} = {\alpha_R} + {\beta_R}({t_i} - c) + {\delta_R} \cdot X + {\varepsilon_{i,R}}$$

If the error satisfies the null expectation assumption, the LATE can be written as: 5 $L A T E = \lim_{t \to c^{+}} E (Y | t = c) - \lim_{t \to c^{-}} E (Y | t = c) = α_{R} - α_{L}, c - h \leq t \leq c + h$ $$LATE = \mathop {\lim }\limits_{t \to {c^ + }} E(Y|t = c) - \mathop {\lim }\limits_{t \to {c^ - }} E(Y|t = c) = {\alpha_R} - {\alpha_L}\:,\:c - h \le t \le c + h$$

However, the relationship on both sides of the breakpoint is not always linear, if the two are nonlinear [30] the above method will produce biased estimates, so by adding a polynomial function f(t_i − c) of the optimization strategy variable D and the allocation variable t to the model, the overall regression function can be expressed as: 6 $Y_{i} = α + L A T E \cdot D_{i} + f (t_{i} - c) + δ \cdot X + ε_{i}$ $${Y_i} = \alpha + LATE \cdot {D_i} + f({t_i} - c) + \delta \cdot X + {\varepsilon_i}$$

The polynomial function is further expanded according to the specific form to construct the following exact breakpoint regression model to capture the relationship between the optimization of the ABC model of financial structure of universities and the relationship between each of the control variables and the level of performance management of the governmental accounting of the overall budget: 7 $\begin{array}{rcl} Y_{i} & = & α + β_{1} D_{i} + \sum_{p = 1}^{p} β_{2} {(t_{i} - c)}^{p} + \sum_{p = 1}^{p} β_{3} {(t_{i} - c)}^{p} \cdot D_{i} \\ + δ_{1} l o g R F I D_{i} + γ_{1} C O M_{1 i} + γ_{2} C O M_{2 i} + ε_{i} \end{array}$ $$\begin{array}{rcl} {Y_i} &=& \alpha + {\beta_1}{D_i} + \sum\limits_{p = 1}^p {{\beta_2}} {({t_i} - c)^p} + \sum\limits_{p = 1}^p {{\beta_3}} {({t_i} - c)^p} \cdot {D_i} \\ &&+ {\delta_1}logRFI{D_i} + {\gamma_1}CO{M_{1i}} + {\gamma_2}CO{M_{2i}} + {\varepsilon_i} \\ \end{array}$$ 8 $D_{i} = {\begin{matrix} 0, t_{i} < \bar{t} \\ 1, t_{i} \geq \bar{t} \end{matrix}$ $${D_i} = \left\{ {\begin{array}{*{20}{c}} {0,{t_i} < \bar t} \\ {1,{t_i} \ge \bar t} \end{array}} \right.$$

Where Y_i is the evaluation index system for the comprehensive budget performance management level of government accounting. $\bar{t}$ $$\bar t$$ is the start date of the optimization strategy, D_i is the generated optimization strategy dummy variable, which represents the processing effect of the optimization strategy, with a certain date as the breakpoint, taking 0 before the implementation of the optimization strategy, and 1 after the implementation of the optimization strategy. (t_i − c) is the standardization of the date allocation variable, so that the breakpoint is always in the 0 position. The equation contains a polynomial in the assignment variable and an interaction term between the assignment variable and the optimization strategy variable, which represents that the slopes of the fitted curves can be different on each side of the breakpoint, increasing the flexibility of the model. This model is next solved using a variety of parameter estimation methods.

3

Research design on the impact of financial structure optimization on budget performance

3.1

Problem Elicitation

The ABC model optimization of financial structure of colleges and universities runs through the whole process of government budget management. Generally speaking, the optimization of the ABC model of financial structure of Chinese colleges and universities is mainly carried out in four aspects. 1)

Decision-making, the reform content corresponds to the newly introduced major decisions, projects, prior performance assessment, enhance the performance constraints of the project process, improve the quality of the project from the source, constraints on the government’s words, actions, and results are consistent, and strengthen the government’s sense of responsibility.

2)

Preparation, the reform focuses on budgeting for projects that meet the requirements of the financial sector and the standardized format, realizing full coverage of budget items and overall departmental performance objectives, and the efficient and smooth implementation of the preparation project, which helps to enhance the credibility of the government.

3)

the implementation of the link, the reform is mainly in accordance with the “two up, two down” procedure, the department from the bottom up to declare the budget project, the NPC from the top down to the official budget and performance targets, emphasizing the combination of daily monitoring and key monitoring, in addition to regular monitoring of routine projects, the individual impact of the project, the focus of the attention of the project for key performance monitoring, timely corrections and improvements to ensure that the performance targets are completed in a quality and quantity to enhance the implementation of the government. In addition to regular monitoring of regular projects, key performance monitoring is carried out for individual projects with high impact and concern, so as to make timely corrections and improvements, ensure that the performance targets are accomplished in terms of quality and quantity, and enhance the government’s executive power.

4)

Completion of the link, the reform focuses on optimizing the performance evaluation management process, forming an interaction between the budget department’s performance self-evaluation and the financial department’s performance evaluation, approving next year’s budget funds with the evaluation results, and incentivizing government departments and other budgetary units to improve their own efficiency and work effectiveness.

How to improve the quality and efficiency of government work and achieve more with less cost has become a key concern of the Government.

3.2

Theoretical analysis and research hypotheses

Under the implementation of ABC model optimization of college financial structure oriented to ABC model optimization of college financial structure, the government began to optimize its organizational structure and improve its efficiency. Research on the relationship between ABC model optimization of university financial structure and government shows that budget performance management oriented to ABC model optimization in university financial structure can not only improve government service efficiency and promote the improvement of social public service level, but also improve government organizational efficiency by having an effect on the internal management of various departments and institutions. Observed from the synergistic perspective of the two, budget performance management is a necessary initiative to enhance the effectiveness of government governance. In addition, the ABC model of university financial structure optimizes the supervision of government decision-making process and administrative behavior, which in turn improves the administrative performance of the government. Thus, it can be seen that the ABC model optimization of university financial structure plays a certain role in improving the administrative performance of local governments. Based on the above analysis, this paper puts forward the hypothesis H1: the optimization of ABC model in the financial structure of colleges and universities helps to improve the administrative performance of the government.

3.3

Variable Selection and Definition

1)

Explained variables

The explanatory variable selected in this paper is local government administrative performance (GOV_i), which is mainly considered from the perspective of administrative scale, and the constructive measurement formula is as follows: 9 $G O V_{i t} = \ln (\frac{p o p u l a_{i t}}{p u b e m p_{i t}} \times n u m c i t y_{i t})$ $$GO{V_{it}} = \ln \left( {\frac{{popul{a_{it}}}}{{pubem{p_{it}}}} \times numcit{y_{it}}} \right)$$

Where popula_i indicates the total population of city i at the end of year t (10,000 people), pubemp_i indicates the number of people employed in public administration and social organizations at the end of city i at the end of year t (10,000 people), numcity_ii indicates the total number of counties and districts under the jurisdiction of city i at the end of year t, and the larger the GOV_i indicates the better the administrative performance of the regional government.

In addition, because the administrative performance of local government from the perspective of administrative scale does not fully reflect the work performance, efficiency and effectiveness of public officials in government departments, this paper simultaneously constructs indicators reflecting the service quality of public officials from the following three indicators.

(1)

Budget request increase (YSSP)

The increase in budget application refers to the growth rate of the number of budget applications submitted by the provincial government to the National People’s Congress at the beginning of the year. The number of budget requests refers to the number of budgets of provinces in the final statement of general public budget revenue and expenditure, including the budgets for recurrent expenditures such as general public service expenditures and diplomatic expenditures, but excluding non-recurrent expenditures such as disasters. The general public budget revenues and expenditures are used within the scope of the law after being considered and approved by the local people’s congresses, with the highest degree of standardization, making the data more informative and comparable. The calculation formula is: 10 $\begin{array}{l} Budget request increase \\ = \frac{Number of budget requests this year - Number of budget requests for the previous year}{Number of budget requests for the previous year} \times 100 % \end{array}$ $$\begin{array}{l} {\text{Budget request increase}} \\ = \frac{{{\text{Number of budget requests this year - Number of budget requests for the previous year}}}}{{{\text{Number of budget requests for the previous year}}}} \times 100\% \\ \end{array}$$

(2)

Fiscal Transparency (FT)

Fiscal Transparency (FT) mainly measures the extent to which information on government financial activities is publicized.

(3)

People’s livelihood expenditure level (PGS)

People’s livelihood expenditure level measures the government’s work on livelihood protection. People’s livelihood expenditure refers to the finalized expenditures of financial departments at all levels in accordance with their functions, which are used to establish social security, employment, education, medical care and other aspects related to the public interest that cover urban and rural residents. The calculation formula is: 11 ${Expenditure level for people}^{'} s livelihood = \frac{{Expenditure on people people}^{'} s livelihood}{General public budget expenditure} \times 100 %$ $${\text{Expenditure level for people's livelihood}} = \frac{{{\text{Expenditure on people's livelihood}}}}{{{\text{General public budget expenditure}}}} \times 100\%$$

2)

The explanatory variables

Constructing the optimization strategy of budget performance management to implement the dummy variable ABC. the optimization of the ABC model of college financial structure basically starts from the performance appraisal of project expenditures, and later gradually joins the departmental budget performance management, because the research object of this paper is the administrative performance of the local government, so the starting time taken combines the substantive content of the relevant documents and the provinces in the optimization of the ABC model of college financial structure. The actual advancement in terms of judgment, to the beginning of the ABC model optimization of college financial structure rises to the local government level of specific initiatives, on behalf of the local government began to seriously implement the ABC model optimization strategy of college financial structure.

3)

Control Variables

In order to avoid the accuracy of the estimation results of this paper being affected by other factors, with reference to the existing relevant studies, the influence of economic growth, population density, industrial structure, urbanization level, financial level, and living standard of residents is controlled here. The specifics of the definition of each variable are shown in Table 1.

Table 1.

Variable definition

Type variable	Name	Variable symbol
Explained variable	Government performance	GOV
Interpretation variable	ABC model structure optimization	ABC
Control variable	Economic growth	Pergdp
	Population density	Density
	Industrial structure	Indstru
	Level of urbanization	Urban
	Financial level	Finan
	Living standard	Living

This paper selects data from 31 provinces, autonomous regions, municipalities and 200 prefecture-level cities in China for the years 2005-2021 as a sample. In the data used in this paper, the time of implementation of ABC model optimization of university financial structure in each province is collected manually through the websites of the Ministry of Finance and local finance departments. The number of people employed in public administration and social organizations in each city is from the China Urban Statistical Yearbook, and the expenditure budget data are from China’s Ministry of Finance and the State Administration of Taxation. Data on social security and employment expenditures, health care and family planning expenditures, education expenditures, and science expenditures were obtained from the Macroeconomic Database. The rest of the data are from the China Urban Statistical Yearbook and provincial statistical yearbooks.

4

Empirical evidence of the impact of financial structure optimization on budget performance

4.1

Descriptive statistics

In order to study the impact of the ABC model optimization of university financial structure on the overall budget performance of government accounting, relevant descriptive statistical analyses are conducted for each variable before regression tests are conducted. In this paper, descriptive statistics are carried out on several variables such as government administrative performance (GOV), ABC model structure optimization of university finance, economic growth (Pergdp), population density (Density), industrial structure (Indstru), urbanization level (Urban), financial level (Finan), and residents’ living standard (Living). The specific analysis is shown in Table 2.

Table 2.

Variable descriptive statistics

Variable	N	mean	p50	sd	min	max
GOV	400	6.954	4.451	0.689	3.458	7.598
ABC	400	0.036	0.038	0.24	-0.99	1.653
Pergdp	400	9.902	9.451	4.595	13.056	9.134
Density	400	422.757	421.561	5	2707	184
Indstru	400	2.317	2.346	1.112	11.553	1.939
Urban	400	0.352	0.456	0.014	2.774	0.201
Finan	400	5620.818	5214.451	140.858	140000	926.556
Living	400	15.329	15.621	10.388	19.737	14.402

4.2

Tests for covariance

The results of covariance analysis are shown in Table 3, when analyzed by variance inflation factor, generally when the tolerance (1/VIF) is smaller, then it means that there is a higher probability of multicollinearity between the variables, and the tolerance is less than 0.1 is considered to be a problem of multicollinearity between the variables. It can also be measured by variance inflation factor (VIF). Tolerance and Variance Inflation Factor (VIF) are inverse of each other, the larger the VIF, the more serious the existence of multicollinearity between the variables, and if the VIF is greater than 10, it is considered that there is a serious problem of multicollinearity between the variables. The VIF are less than 2.5, and the smallest value of 1/VIF is 0.469, which proves that there is a low likelihood of the existence of the problem of multicollinearity.

Table 3.

Linear test

Variable	VIF	1/VIF
ABC	1.278	0.782
Pergdp	1.686	0.593
Density	1.481	0.675
Indstru	1.313	0.762
Urban	2.134	0.469
Finan	1.694	0.590
Living	1.587	0.630
Mean VIF	1.681	0.595

4.3

Analysis of OLS regression results

In order to compare with the optimization strategy effects analyzed by the breakpoint regression design, this paper also reports the regression results under ordinary least squares OLS, and Table 4 shows the results of the OLS regression on the sample data. Columns (1) and (2) show the effects of the optimization of the ABC model in the financial structure of universities on the increase in budget requests, financial transparency, and the level of livelihood expenditures after the inclusion of control variables, respectively. Column (1) shows that the regression coefficient of budget request increase is negative at 1% significance level. The optimization of the ABC model in the financial structure of colleges and universities makes the budget request increase to be -15.23%, i.e., a decrease of 15.23%, while other variables are kept constant. From column (2), it can be seen that the coefficient of fiscal transparency at 1% significance level is 15.28 after the optimization of ABC model in the financial structure of universities, which makes the score of fiscal transparency increase by 15.28 with other variables kept constant. From column (3), it can be seen that the optimization of ABC model in the financial structure of universities increases the level of livelihood expenditures with a The regression coefficient is 1.035, and the optimization of the ABC model in the financial structure of colleges and universities increases the level of livelihood expenditure by 1.035% while other variables are kept constant, but it does not pass the significance test. However, this regression is not entirely accurate, as can be seen from the last row of Table 4 where the R2 obtained from this regression is 61.9%, 36.4% and 28.7% respectively, suggesting that there is a possibility of an omission of the key variables, which results in biased regression coefficients being obtained. Consequently, the next test was conducted using breakpoint regression to address the issue of endogeneity. (The standard errors calculated in parentheses there are cluster robust standard errors. (* denotes p<0.1,** denotes p<0.05,*** denotes p<0.01, which means that the estimates are significant at 10%, 5% and 1% significance levels, respectively.)

Table 4.

Ols regression results

	(1) YSSP	(2) FT	(3) PGS
D	-15.23^***	15.28^***	1.035
Coefficient	(0.908)	(2.294)	(0.679)
Whether to add control variables	yes	yes	yes
α	32.56^***(2.087)	19.15^***(4.613)	31.549^***(6.456)
N	400	400	400
R²	0.619	0.364	0.287

4.4

Analysis of breakpoint regression results

Through OLS analysis, we are able to get the initial impact of optimizing the implementation of the ABC model in the financial structure of the university. Next, the data will be further examined using breakpoint regression.

First of all, to use breakpoint regression we have to test the validity of the explanatory variables. The explanatory variable in the paper is time, the optimization of the ABC model in the financial structure of colleges and universities, as well as the time of optimization are not controlled by the provinces and are not announced in advance, the provincial departments can not be prepared in advance, so it meets the randomness. Secondly, the premise of using the breakpoint regression method is that the explanatory variables have clear breakpoints. Figure 1-Figure 3 are scattered plots of the increase in budget application, fiscal transparency and the level of people’s livelihood expenditure before and after the breakpoint, and it can be seen through graphical analysis that the vertical dashed line is the boundary, divided into the left side of the breakpoint and the right side of the breakpoint, the left side of the breakpoint of the budget application increase is significantly higher than the right side of the breakpoint, and there is no overlap in the confidence interval, the right side of the breakpoint of fiscal transparency and the level of people’s livelihood expenditure is significantly higher than the left side of the breakpoint, and the confidence interval overlaps less. It shows that there is a significant gap between the left and right sides of the breakpoint, which also proves that there is a breakpoint present here. In conclusion, it proves the feasibility of the breakpoint regression method in this study.

Breakpoint regression design is categorized into parametric and non-parametric estimation methods, which have their own advantages and disadvantages. Parametric estimation is convenient and efficient. However, if the model is not set appropriately, it may generate setting errors. Non-parametric estimation does not need to design a specific regression model, and the obtained regression coefficients are more robust, but it requires a relatively large sample capacity. Combined with the sample selection, this paper focuses on breakpoint regression analysis by parameter estimation, and the regression results of nonparametric estimation are supplemented in the robustness test. The parameter estimation bandwidth is set manually, with reference to the existing literature, the bandwidth in this paper is selected as three years before and after the breakpoint (including 2011), i.e., a total of seven years from 2010 to 2016, with a total of 400 samples, and based on the theory that the estimation of the low-order polynomials is better than that of the high-order polynomials, the order of the high-order terms in the model is only selected up to the third order. The parameter estimation regression results are shown in Table 5.

Table 5.

Parameters are estimated to return

	(1)YSSP	(2)FT	(3)PGS	(1)YSSP	(2)FT	(3)PGS
D	-33.84^***	-5.645	0.864	-33.49^***	-6.184	0.885
Coefficient	(3.5210)	(10.654)	(1.1564)	(3.5642)	(10.5486)	(1.1547)
Whether to add control variables	no	no	no	yes	yes	yes
α	49.54^***	38.546^***	37.45^***	52.45^***	34.512^**	37.15^***
N	400	400	400	400	400	400

Columns (1) to (3) show the regression results for the increase in budget requests, fiscal transparency and the level of livelihood expenditures, respectively, while columns (4) to (6) reflect the results after the inclusion of control variables, respectively. From column (1), it can be seen that the regression coefficient of the effect of the optimization of ABC model in the financial structure of colleges and universities on the increase of budget request is -33.84 at 1% significance level, indicating that the optimization of ABC model in the financial structure of colleges and universities has led to an increase of -33.84%, that is, a decrease of 33.84%. The optimization of the ABC model in the financial structure of universities requires that budget requests are made in the direction of the estimated behavior to control the savings and reduce the blind budget requests of government departments and units. The regression results show that the optimization of the behavioral module of the ABC model in the financial structure of universities and colleges does significantly slow down the growth rate of budget requests at the provincial level, which is in line with the expectations and research hypotheses. The regression coefficient of column (4) is -33.49, which is very close to the data in column (1), indicating that the inclusion of control variables does not have much impact on the estimated value of the effect of optimization strategy. Combined with the trend in Figure 1, it is again found that the decrease in the increase in budget requests is an effect in the short term. Since the budget performance management system is not yet perfect, it may weaken the optimization strategy effect in actual implementation.

Column (2) shows that the optimization of the ABC model in the financial structure of universities does not have a significant effect on the improvement of provincial fiscal transparency, with a regression coefficient of -5.645, the regression result is insignificant, and the optimization strategy has a limited effect on the improvement of fiscal transparency. After adding control variables, the regression coefficient of column (5) is -6.185, still not significant. It can be seen that the level of economic development also has no significant effect on the implementation of the budget performance optimization strategy, and the expected effect of the implementation of the optimization strategy is not achieved.

Column (3) shows that the regression coefficients of the level of livelihood expenditures with and without control variables are 0.864 and 0.885 respectively, both of which are insignificant, indicating that the optimization of ABC model in the financial structure of colleges and universities does not have a significant impact on the level of livelihood expenditures, and does not achieve the expected effect of the implementation of optimization strategies. After the optimization of the ABC model in the financial structure of colleges and universities, the local government focuses on the efficiency of the use of funds, and the saved funds are not used more in the area of livelihood expenditure. Even if some of the financial funds are saved by reducing the budget application, the investment in livelihood expenditures has not increased significantly, and the reasons need to be further found.

In the impact on the budget performance of local government accounting, the budget application increase has the most significant impact on it, and this paper confirms that the growth rate of budget application at the provincial level slows down significantly after the optimization of the behavioral module of the ABC model in the financial structure of colleges and universities, and therefore also verifies the previous hypothesis H1: the optimization of the ABC model in the financial structure of colleges and universities can help to improve the performance of the administration of the local government.

4.5

Robustness Tests

The results of the robustness test are shown in Table 6, in column (1), this paper increases the control variable of the classification of colleges and universities, which is 1 if the college or university is a first-class university construction college or university, and 0 if the college or university is a first-class academic discipline construction college or university. The regression coefficient of the explanatory variable ABC model optimization of the financial structure of colleges and universities in the regression results is -0.487, which is significant at the level of 1%, and it supports the hypothesis H1.

Table 6.

Robustness test results

VARIABLES	(1)GOV	(2)GOV	(3)GOV
ABC	-0.487^***	-0.305^***	-0.316^***
Coefficient	-4.66	-2.91	-3.08
Pergdp	-0.035	-0.112	-0.045
Coefficient	-0.25	-1.34	-0.34
Density	0.000		-0.000
Coefficient	0.35		-1.11
Indstru	-0.006	-0.012	-0.005
Coefficient	-0.52	-1.13	-0.23
Urban	0.000^***	0.000^***	0.000^***
Coefficient	8.06	13.81	12.55
Finan	0.311^***	0.052	0.058
Coefficient	6.21	0.75	0.78
Living	0.035	0.106^*	0.155^**
Coefficient	0.57	1.81	2.13
Y	0.744^***
Coefficient	12.54
avg Pergdp		0.000^**
Coefficient		2.33
Constant	-2.195^***	-2.594^***	-2.744^***
Coefficient	-12.95	-13.54	-10.87
Observations	400	400	350
R-squared	0.679	0.464	0.461
year	YES	YES	YES
r2_a	0.65	0.441	0.434
F	60.22	55.48	40.31

In (2), this paper replaces the control variables, replacing the local economic growth Pergdp with Per capita Pergdp, i.e., the variable avgPergdp. The regression coefficient of the ABC model optimization of the financial structure of colleges and universities in the regression results is -0.305, which is also significant at the 1% level, supporting hypothesis H1.

In (3), the time interval of the sample was adjusted to 2016-2021, a total of six years for the sample data. In the regression results, the regression coefficient of ABC model optimization of financial structure of universities on budget performance is -0.316, which passed the test of significance level of 1%.

4.6

Heterogeneity test

In order to further understand the specific effects of the ABC model optimization reform of university financial structure, this paper explores the heterogeneity effects from the dimensions of institutional environment, government audit, and financial decentralization. Table 7 shows the results of the heterogeneity test.

Table 7.

Heterogeneity test

Variable	Institutional environment		Government audit		Degree of fiscal decentralization
Variable	Low	High	Low	High	Low	High
ABC	-0.225^**	-0.045	-0.987	0.034^***	0.023	-0.185^***
Coefficient	0.111	0.034	0.122	0.042	0.075	0.071
Control variable	control	control	control	control	control	control
Provincial fixation	control	control	control	control	control	control
Vintage fixation	control	control	control	control	control	control
Sample size	400	400	400	400	400	400
R²	0.748	0.911	0.812	0.965	0.914	0.852

This paper examines the effect of institutional environment on the relationship between ABC model optimization of university financial structure and budget performance management. The marketization index is used to reflect the institutional environment. This paper is divided into two groups of high degree of marketization and low degree of marketization according to whether the degree of marketization exceeds the median of the sample. As can be seen from the regression results in Table 7, the regression coefficient of the low degree of marketization group is significant at -0.225 and significant at the 5% level, while the regression coefficient of the high degree of marketization group is -0.045, which is not significant. The regression results indicate that the optimization of the ABC model of financial structure of universities is more capable of promoting budget performance management in regions with poor institutional environment. The possible reason for this is that with the continuous improvement of the institutional environment, the marginal benefit of the ABC model optimization of university financial structure is decreasing. Regions with better institutional environments are inherently more efficient in the allocation of financial resources, and the same governance role in these regions may just be the icing on the cake, with lower marginal returns. While in the region with poorer institutional environment, the governance role of the ABC model optimization of college financial structure may play the role of sending charcoal in the snow, with higher marginal returns.

This paper further examines the impact of government audits on the relationship between the ABC model optimization of university financial structure and budget performance management. The intensity of government audits in the locality is expressed using the proportion of current financial expenditures that should be surrendered from the problems identified by the audits. The paper is divided into two groups, weak government audit and strong government audit, based on whether the government audit intensity exceeds the median of the sample. The regression coefficient of the weak government audit group is significantly negative and significant at the 1% level, and the regression coefficient of the strong government audit group is not significant. The regression results indicate that budget performance management reforms are more able to promote budget performance management in areas with weak government audits. The possible reason is that, as the “immune system” of the national governance system, government audit is an important tool for local governments to improve the efficiency of financial expenditures and ease the performance management of local budgets. The mitigating effects of government auditing and budget performance management on local government budget performance management complement each other and play a governance effect together. When the government audit is weak, the governance effect of budget performance management will be more obvious.

This paper further examines the impact of fiscal decentralization on the relationship between ABC model optimization of university financial structure and budget performance management. Measuring the degree of fiscal decentralization from the perspective of fiscal revenue, this paper is divided into two groups of low degree of fiscal decentralization and high degree of fiscal decentralization according to whether the degree of fiscal decentralization exceeds the median of the sample. The regression coefficients for the low degree of fiscal decentralization group are not significant, and the regression coefficients for the high degree of fiscal decentralization group are significantly negative and significant at the 1% level. The regression results indicate that the optimization of the ABC model of financial structure of higher education institutions is more capable of promoting budget performance management in regions with high degree of fiscal decentralization. The possible reason is that local governments with a high degree of fiscal decentralization have more control and decision-making power over the financial and economic resources of their jurisdictions, and the motivation of “promotion tournament” is stronger, so officials have stronger incentives to implement the ABC model optimization of university financial structure.

5

Conclusion

This paper investigates the effect of the optimization of the ABC model of financial structure of colleges and universities on the overall budget performance of government accounting by constructing a breakpoint regression analysis model.

Using breakpoint regression analysis, the implementation effect of advancing the optimization of the ABC model of financial structure in colleges and universities is examined. After the optimization of ABC model of financial structure of Chinese universities, the decrease of budget application is 15.23%, which indicates that the optimization strategy of ABC model of financial structure of universities has a significant effect on the reduction of the increase of budget application, and the government starts to save budget expenditure comprehensively, and declares budgets more rationally with behavioral orientation. It shows that the effect of the ABC model optimization strategy of university financial structure has been shown, but its effect on the improvement of two indicators, namely, the level of people’s livelihood expenditure and financial transparency, is not significant. It is necessary to further find the deep-seated reasons, but overall the ABC model optimization of university financial structure has an enhancing effect on government administrative performance.

In order to further verify the experimental results of the breakpoint regression analysis of this paper, the robustness test is carried out by replacing the control variable of economic growth Pergdp and the time interval of the samples, etc. The results of the robustness test once again prove the reliability of the empirical results of this paper.

The results of heterogeneity analysis found that the optimization of the ABC model of university financial structure is more effective for government budget performance management in regions with poorer institutional environment, weaker government audit and higher degree of fiscal decentralization. It shows that improving the market institutional environment, exerting the governance effect of government auditing, and maintaining a unified fiscal decentralization system are greatly beneficial to government performance management.

Language:: English

Publication timeframe:: 1 times per year
Journal Subjects:: Life Sciences, Life Sciences, other, Mathematics, Applied Mathematics, General Mathematics, Physics, Physics, other

Journal RSS Feed

Analysis of ABC Model for Optimizing Financial Structure of Universities in Government Accounting Comprehensive Budget Performance Management

Qingquan Huang

Published Online: Sep 26, 2025

Received: Jan 02, 2025

Accepted: Apr 20, 2025

DOI: https://doi.org/10.2478/amns-2025-1076

KeywordsABC model, Breakpoint regression analysis method, OLS regression, Government budget performance management

© 2025 Qingquan Huang, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Keywords
ABC model, Breakpoint regression analysis method, OLS regression, Government budget performance management