Research on the role of AHP- entropy method in the identification and evaluation of China tariff source risk

Abstract

With the rapid development of China’s foreign trade, more and more attention is paid to the quality and efficiency of customs risk management. However, there are some subjective defects in most of the current tax risk assessment methods, which makes the results of risk assessment lack of authenticity and accuracy. Based on this, the identification and evaluation method of China tariff source risk based on AHP- entropy method was put forward in this paper, the problems existing in the current China’s tariff tax risk management method were analyzed on the basis of the brief introduction of the related theories and technologies, the principles and characteristics of AHP- entropy method were described, and the tariff source risk identification index system based on AHP- entropy method was constructed, then the recognition, measurement and evaluation of the sources of risk were carried out. Experimental results show that the tax risk assessment based on AHP-entropy method can effectively improve the accuracy of the evaluation results and the accuracy of the risk response, and improve the quality and efficiency of risk management.

Keywords

Risk management customs tariff AHP-entropy method

1 Introduction

In recent years, the shortages of human resources, supervision, efficiency and other issues have become increasingly prominent, the existing customs supervision mode has been unable to meet the needs of the development of the new situation. How to establish a strict and efficient customs risk management model under the limited manpower and material resources has become the development needs. Foreign related researches started earlier, after decades of development, has formed a relatively mature technologies and theories. Although domestic related researches started late, but in recent years, some progress has been made. Based on the connotation of risk and research status, the theoretical framework of the research of risk management was proposed by the Wangzhong in 2005, and risk analysis model was outlined and summarized, which provided a theoretical basis for the research on risk management of customs [1]. A BP neural network based on Ehrenberg learning algorithm was introduced into the risk assessment of Customs enterprises on the basis of the BP neural network data mining method by Lu Jinqiu in 2006, experimental results shown that the algorithm effectively improved the customs of enterprise risk control level [2]. A corporate risk assessment based on the customs clearance system was proposed by Huang Yamiao in 2009, the method fully played the advantages of classified customs clearance system, and improved the efficiency of customs supervision to a certain extent [3]. Based on the risk management theory, the customs special supervision regional development model and the customs risk analysis, the reasons of risk analysis were carried out by Geng Changjun in 2013, and the countermeasures to solve the risks were put forward [4]. Based on the actual situation of Guangdong customs, the problems existing in the current customs supervision mode was analyzed by Jiang Weiping in 2014, and the idea of promoting risk management in the next step of customs supervision was put forward [5]. Based on the selection of representative index system and the introduction of analytic hierarchy process, the weight of each index was determined by Deng Shaojun in 2015, through the establishment of the hierarchy model, the structure of the judgment matrix, the relative weight and the calculation of the synthetic weight, the reasonable proposals for the further improvement and improvement of export toys inspection and supervision risk assessment were put forward [6]. However, there are some subjective defects in most of the current tax risk assessment methods, which makes the results of risk assessment lack of authenticity and accuracy. Therefore, it is urgent to introduce more scientific and reasonable risk assessment method to make the sources of risk assessment results more accurate and reliable. The emergence of AHP and entropy method not only enriches and perfects the index weight set of methodology system, and provides the reference for the improvement and update of sources of risk assessment methods.

Based on this, the evaluation framework based on AHP and entropy method in the tax risk was proposed in this paper. The remainder of this paper is organized as follows. Section 2 describes the relevant theories and techniques. Section 3 gives the study on the evaluation framework based on AHP and entropy method in the tax risk. Section 4 presents a real experiment to evaluate the method. Conclusion is summarized in Section 5.

2 State of the ART

2.1 Tariff tax risk management

With the rapid development of China’s foreign trade, the contradiction between the limited regulatory resources of the customs and the rapid growth of the regulatory task is more and more prominent, especially the risk of smuggling of illegal, there is a trend of further increase [7]. Foreign trade developed countries such as the United States, Australia, Holland and the United Kingdom and other countries of the customs are implementing risk management, which resolves the contradiction between limited regulatory resources and the growing regulatory task [8]. In the implementation of customs risk management, there are three aspects: one is the effective use of the limited supervision resources, supervision will continue to increase the import and export trade; the second is to provide convenience for the law-abiding import and export enterprises of trade; the third is to reduce the cost of customs management [9]. Risk identification is the foundation of the whole risk management. Therefore, reasonable risk classification is the premise of identifying risk, and it is also needed to make use of various scientific methods and techniques (Fig. 1).

Fig.1

1980–2014 Tariff tax.

Since there is no way to identify all the risks, it is often needed to use a variety of identification methods to improve the accuracy of risk identification [10]. At present, customs risk identification methods often used mainly include the following categories: regulatory process analysis method, business expert survey method, mathematical statistics analysis method, case analysis method, dynamic data monitoring method, the regulatory environment analysis. However, there are still many problems and difficulties in the current customs risk management in China [11]. Mainly in: Overall, the understanding and awareness of risk management and its importance are not deep enough. There is no risk prevention and control mode of scientific system. Unified risk management information platform has not yet formed, the integration of information is lacked, implementation of risk management of the external environment is not perfect [12].

2.2 AHP- entropy method

AHP method is to use the less quantitative information to make the decision-making process of mathematical thinking on the basis of in-depth analysis of complicated decision problem essence, influence factors and its internal relations, so as to provide a simple decision method for mufti-objective, mufti criteria or no structural properties of complex decision problem [13]. Specifically, it divides the relevant elements into objectives, guidelines, programs and other levels. It provides quantitative basis for the analysis, decision making, prediction and control of human’s thinking process by hierarchical and quantitative analysis. The structure and steps of AHP are shown below (Fig. 2):

Structural analytic hierarchy

Fig.2

Hierarchy structure.

The decision-making problem is divided into target layer, criterion layer and index layer. Hierarchical structure diagram as shown below.

Construct two-two comparison discriminant matrix.

For each level of the relative importance of the index score, the 1–9 scale method is used to scale the relative importance of the two-two indicators [14]. The judgment matrix and the 1–9 scaling method are as follows (Table 1).

Consistency test of judgment matrix

Table 1

1–9 scaling method

Serial number	Importance level	C_ij assignment
1	i,j two elements are equally important	1
2	i is slightly more important than j	3
3	i is significantly more important than j	5
4	i is strongly important than j	7
5	i is extremely important than j	9
6	i is not slightly more important than j	1/3
7	i is not significantly more important than j	1/5
8	i is not strongly important than j	1/7
9	i is not extremely important than j	1/9

The consistency of judgment matrix refers to the coordination of each judgment when judging the importance of the index, which does not appear to contradict each other. In the analytic hierarchy process, the following formula is used to check the consistency of the decision maker’s thinking. $CI = \frac{λ_{max} - n}{n - 1}$ (1)

The greater the CI value, the greater the degree of deviation from the complete consistency of the judgment matrix, the lower the CI value, the better the consistency of the judgment matrix.

When the judgment matrix is completely consistent, CI = 0;

When the judgment matrix is satisfied, it is necessary to introduce the average random consistency index RI value of the judgment matrix. For the 1–9 order judgment matrix, RI values in Table 2.

Table 2

RI value

1	2	3	4	5	6	7	8	9
0.00	0.00	0.58	0.90	1.12	1.24	1.32	1.41	1.45

When the order is greater than 2, the consistency of the judgment matrix CI and the same order average random consistency index RI ratio is called the random consistency ratio CR. When CR = CI/RI < 0.10, it can be considered that the judgment matrix has a good consistency, otherwise, it needs to adjust the judgment matrix (Table 3).

Hierarchical single ranking

Table 3

Hierarchical total ranking

Level A	B1	B2	B3	Total ranking
Level B	0.105	0.637	0.258
C1	0.491	0	0.406	0.157
C2	0.232	0.055	0.406	0.164
C3	0.092	0.564	0.094	0.393
C4	0.138	0.118	0.094	0.113
C5	0.046	0.263	0	0.172

The maximum eigenvalue and the corresponding eigenvalues are obtained by iterative method on the computer. In this paper, the calculation steps of the method of calculating the maximal eigenvalue of the matrix and the corresponding characteristic vector are given:

The product of each line element of a matrix of computation M_i is: $M_{i} = \prod_{j = 1}^{n} a_{ij}$ (2)

Calculating n root mean square of M_i: $\bar{W_{i}} = \sqrt[n]{M_{i}}$ (3)

The vector $\bar{W} = {[\bar{W_{1}}, \bar{W_{2}}, \dots, \bar{W_{n}}]}^{T}$ was normalized: $W_{i} = \frac{\bar{W_{i}}}{\sum_{j = 1}^{n} \bar{W_{j}}}$ (4)

W = [W₁, W₂, …, W_n] ^T is the desired feature vector.

The largest eigenvalue of the matrix is: $λ_{max} = \sum_{i = 1}^{n} \frac{{(AW)}_{i}}{{nW}_{i}}$ (5)

Among them, (AW) _i represents the first i element of vector AW.

For the judgment matrix A, the calculation result is:

$\begin{matrix} W & = & [\begin{matrix} 0.105 \\ 0.637 \\ 0.258 \end{matrix}], λ_{max} = 3.308, \\ CI = 0.019, RI = 0.58, CR = 0.033 \end{matrix}$ (6)

For the judgment matrix B₁, the calculation result is:

$\begin{matrix} W & = & [\begin{matrix} 0.491 \\ 0.232 \\ \begin{matrix} 0.092 \\ 0.138 \\ 0.046 \end{matrix} \end{matrix}], λ_{max} = 5.126, \\ CI = 0.032, RI = 1.12, CR = 0.028 \end{matrix}$ (7)

For the judgment matrix B₂, the calculation result is:

$\begin{matrix} W & = & [\begin{matrix} 0.491 \\ 0.232 \\ \begin{matrix} 0.092 \\ 0.138 \end{matrix} \end{matrix}], λ_{max} = 4.117, \\ CI = 0.039, RI = 0.90, CR = 0.043 \end{matrix}$ (8)

For the judgment matrix B₃, the calculation result is:

$\begin{matrix} W & = & [\begin{matrix} 0.406 \\ 0.406 \\ \begin{matrix} 0.094 \\ 0.094 \end{matrix} \end{matrix}], λ_{max} = 4, \\ CI = 0, RI = 0.90, CR = 0 \end{matrix}$ (9)

2.3 Entropy method

Entropy is a measure of uncertainty of system state. The evaluation of all the schemes in the inherent information can be used, the information entropy of each index is obtained through the method of entropy, the smaller the information entropy, the lower the degree of disorder information, and the utility of the information value is greater, the weights of the indexes is also greater. The calculation of the difference coefficient of entropy method includes the following steps [15].

Firstly, the evaluation matrix is established. Setting there are m objects to be evaluated, n evaluation indicators of the assessment, according to the evaluation of the object of the sample observation data to get the initial evaluation matrix $R^{'} = {(r_{ij}^{'})}_{m \times n}$ . Where $r_{ij}^{'}$ is the state value of the i object on the first j.

Secondly, data standardization. In the evaluation system, the dimension of the index, the unit of dimension and the positive and negative orientation of the index may be different, so when the system is evaluated, it is necessary to carry out the standardization of the data first, and get the standard state matrix R = (r_ij) _m×n. The standardization formula is as follows: $r_{ij} = {\begin{matrix} \frac{r_{ij}^{'} - min_{i} (r_{ij}^{'})}{max_{i} (r_{ij}^{'}) - min_{i} (r_{ij}^{'})} max_{i} (r_{ij}^{'}) \neq min_{i} (r_{ij}^{'}) \\ 1 max_{i} (r_{ij}^{'}) \neq min_{i} (r_{ij}^{'}) \end{matrix}$ (10)

Among them, R_ij is the first j index of the second evaluation object, $min_{i} (r_{ij}^{'})$ is the minimum value of the j index, and $max_{i} (r_{ij}^{'})$ is the maximum value of the j index.

Thirdly, data normalization processing. the standard matrix is got. The normalization formula is as follows: $θ_{ij} = r_{ij} / \sum_{i = 1}^{m} r_{ij} (0 \leq θ_{ij} \leq 1)$ (11)

Fourth, calculating the j index of entropy. The calculation formula is as follows:

$\begin{matrix} H_{j} & = & - k \sum_{i = 1}^{m} θ_{ij} 1 n θ_{ij}, j = 1, 2, \dots, n \\ (k = 1 / 1 nm) \end{matrix}$ (12)

Fifth, calculating the difference coefficient of the first j index g_j. The calculation formula is as follows: $g_{j} = 1 - H_{j}, j = 1, 2, \dots, n$ (13)

3 Methodology

3.1 Sources of risk assessment based on AHP-Entropy method

In this paper, the entropy difference coefficient of hierarchical analysis weighting method is modified to obtain the optimal combination weight, calculation formula is as follows: $w_{i} = w_{i} g_{i} / \sum_{j = 1}^{n} w_{i} g_{i}, i = 1, 2, \dots, n$ (14)

Where w_i is the combination weight of the AHP- entropy method, W_i is the weight of AHP, g_i is the coefficient difference entropy method. Based on this, this paper presents a method to evaluate the risk sources based on the AHP-Entropy method, the steps are as follows:

Clearing tax risk assessment target

The goal of the sources of risk assessment is to carry out a measure of accurate and objective evaluation of the sources of risk, effective recognition through the tax risk identification index system, and help the tax department to find the manner taxpayers tax risk in time, guarantee the authenticity and reliability of the results of the risk assessment, so as to constantly improve the high tax authorities tax source monitoring ability, and prevent and reduce the loss of tax revenue.

Constructing the tax risk identification index system

Sources of risk identification index system is a group consisting of a number of interrelated and influence each other tax related indicators characteristics index collection system, which can comprehensively reflect the rules and characteristics of the sources of risk. The accuracy and pertinence of the selection of the sources of risk identification index system will directly affect the ability to effectively identify the sources of risk. In the construction of tax risk identification index system, it is need to follow the following principles: one is quantitative principles, it is needed to on the basis of the current statistical methods, the data can be obtained, which are easy to understand and can be used for quantitative analysis. Two is the comparison principle, the analysis index system should include horizontal comparison index and longitudinal comparison index, which not only can meet the same region or the same industry horizontal comparison of different taxpayers, but can meet the longitudinal comparison of the same enterprise, so as to ensure that the evaluation results can reflect and reveal the problems. Three is the principle of feasibility, it is needed to ensure the selection of evaluation indicators easy to carry on the data collection and analysis, and has a wide range of feasibility.

Tax risk identification index system weight set

Due to the complexity and diversity of the sources of risk, the construction of tax risk identification index system of risk assessment of importance and emphasis is not the same. So it is necessary to give a certain weight to the sources of risk identification index system, so as to reflect the importance of sources of risk indicators to ensure tax risk comprehensive evaluation more substantial and accurate. This paper uses a AHP entropy method which combines subjective and objective weight identification index system.

Sources of risk identification

Based on the different dimensions of risk identification, risk identification of sources of tax revenue is to use the tax risk identification index system to fully scan all taxpayers from different levels or different perspectives for tax risk areas. Through the lateral identification, vertical identification, correlation recognition, tax payment ability and key link monitoring to find and lock the risk characteristics of the taxpayer’s process.

The sources of risk measure

Sources of risk measurement is the process of the tax department carries on the quantification analysis to the type of risk, tax risk points of the key index deviates from the standard value degree, resulting in tax loss severity. For different sources of risk identification index, it is needed to use different risk measure method, all kinds of sources of risk are accurately measured, so as to obtain each tax risk risk measure index.

Comprehensive tax risk assessment

Comprehensive tax risk assessment includes tax risk index and risk rating and synthetic ranking. Risk assessment departments consider sources of risk of all factors, and according to all households separately taxpayer risk measure to use synthetic risk model to synthesize the single sources of risk measure value and index weight, so as to form a weighting form sources of risk index. Then according to the prior to determine the sources of risk classification standards, and carry out the risk rank.

The tax risk assessment results push

Sources of risk assessment results push is that the risk assessment department according to the level of taxpayer risk level to push the risk assessment results to different risk response departments, so that the department can deal with risk the difference of risk coping. The purpose of the risk assessment results push is to implement the risk effectively respond, so that the tax source management department can take the optimization of tax services to deal with the risk, and let taxpayers voluntarily make up the taxes.

3.2 China tariff source identification index system of risk identification

The specific performance of China’s customs tariff source risk was analyzed in this paper, the index system of duty tax risk identification is shown in Fig. 3.

Fig.3

Tariff tax risk identification index system.

The tariff tax risk form

China customs risk can be classified as price fraud risk, violation of the customs supervision provisions of risk, the data statistics risk and other random risk.

Risk of price fraud

A common feature of price fraud risk is to use various methods (low reported prices, less reported clinch a deal the price composition part, Pseudonym, contract decomposition, sales commission) to declare the transaction price of the goods, so as to achieve the purpose of lower paid price and less tax. The direct loss of such risks is the loss of taxes.

Violation of the provisions of the customs supervision risk

The common characteristics of the violation of the provisions of the customs supervision risk is the violation of the provisions of the customs supervision. The regulations of the customs supervision and control shall refer to the foreign trade control policy and management system based on the law enforcement activities. Such risk shall be divided into import and export licensing system, quota management, intellectual property protection, anti-dumping and anti subsidy, radio communication equipment into the exit and prints entry and exit management sub class in accordance with the occurrence and damage of the field.

Data statistics risk

Data statistical risk refers the harm that the data acquisition inaccurate and scientific data processing cause the customs import and export goods trade statistics, customs operation statistics and management of customs required a variety of business application system inaccurate and not timely.

3.3 Sources of risk identification dimensions

Based on the reference and use of key indicators analysis recognition method, and the combination of the characteristics of China tariff revenue, this paper identifies the tariff tax risk respectively from the horizontal, vertical, relevance, tax capacity estimation and key tax source monitoring and multiple dimensions.

Cross comparison recognition

Horizontal comparison is a comparison among the different objects of the same kind of thing, under normal circumstances, standard of the same kind of things is unified, through the comparison of the standard of similar things, value, quality and size for the same thing, the things that reach the standard can be identified, so as to provide references for decision-making. In the tariff source of the risk identification process, the same goods or goods is homogeneous, through the comparison to identify the abnormal behavior of the taxpayers to pay taxes with the standard values.

Longitudinal comparison identification

If the customs tax revenue fluctuates greatly in a certain period of time, then it shows the risk of tax payment is abnormal, and there is the risk of low paid or unpaid taxes.

Identification of Association

From the perspective of tax, customs tax often involves a number of tax declare and pay, the tax exists certain relevance, the various related tax tax basis must be the same, through the related tax tax basis by matching analysis, it can be found that the taxpayer is no less paid or unpaid taxes.

Estimation of tax payment ability

The vertical fairness of tax revenue refers to the taxpayers who have different tax ability to pay the corresponding tax, in order to realize the maximization of the taxpayer’s ability to pay tax. In the tariff source risk assessment, through the comparison of the taxpayer a tax payable tax and the actual payment of tax revenues to identify taxpayers pay less tax revenue risk.

Key link monitoring

In addition to risk identification from the point of view, the tax department should use effective monitoring methods and control means to monitor the key link, timely prevent the link of tax related risks, and reduce the loss of tax revenue.

4 Result analysis and discussion

In order to verify the effectiveness of the method, the simulation experiment is carried out (Table 4).

Table 4
China Customs business situation from 2010 to 2015 Unit (100 million yuan)

Particular year Penalty and Confiscatory Income Storage system Total tax relief Import and export declaration form Smuggling cases Smuggling case value

2010 19.5 12518.3 390.0 5723 13076 269.8

2011 22.7 16142.1 420.3 5903 12445 214.6

2012 25.1 17639.1 450.6 6042 17021 354.7

2013 26.2 19467.8 470.1 7903 19801 400.6

2014 28.9 20056.9 509.2 8530 25436 440.4

2015 30.3 24064.1 659.0 9758 28963 509.9

Particular year	Penalty and Confiscatory Income	Storage system	Total tax relief	Import and export declaration form	Smuggling cases	Smuggling case value
2010	19.5	12518.3	390.0	5723	13076	269.8
2011	22.7	16142.1	420.3	5903	12445	214.6
2012	25.1	17639.1	450.6	6042	17021	354.7
2013	26.2	19467.8	470.1	7903	19801	400.6
2014	28.9	20056.9	509.2	8530	25436	440.4
2015	30.3	24064.1	659.0	9758	28963	509.9

The results show that this method improves the risk identification strategy and the ability of risk identification, reduces administrative costs, improves the effectiveness of supervision, and promotes the healthy development of foreign trade. the tax risk assessment based on AHP-entropy method can effectively improve the accuracy of the evaluation results and the accuracy of the risk response, and improve the quality and efficiency of risk management.

5 Conclusion

In order to facilitate the supervision of limited resources, strengthen the regulation and compliance with the requirements of trade, China’s customs must implement risk management. Based on this, this paper uses AHP-entropy method to carry out the weights to index system, and constructs the tariff revenue risk identification index system based on AHP-entropy method, so as to carry out the recognition, measurement and evaluation of the sources of risk. This method takes the combination of subjective and objective way, changes the sources of revenue risk assessment on the status of subjective evaluation and makes the risk assessment result more objective, accurate and reliable. In addition, the AHP-entropy method is used for identifying index system weight, through the hierarchical analysis method to determine the subjective weights, then the difference coefficient of entropy method is used to modify the subjective weight, so that the weight of each index not only can reflect the opinions, but can reflect the characteristics of indexes. The results show that this method improves the risk identification strategy and the ability of risk identification, reduces administrative costs, improves the effectiveness of supervision, and promotes the healthy development of foreign trade.

Footnotes

Acknowledgments

This paper was supported by Shanghai First-class Academic Discipline Project, Project Number: S1201YLXK.

References

Jin-Qiu

L.U.

, Application research on customs risk-management based on artificial neural networks, Computer Engineering & Applications42(27) (2006), 208–211.

Boehm

B.W.

, Software risk management: Principles and practices, IEEE Software8(1) (1991), 32–41.

Acharya

, Thirty years of tax reform in India, Economic & Political Weekly40(20) (2005), 2061–2070.

Q.S.

, Shi

J.G.

and Zhang

Y.G.

, Evaluation on the development of shanghai suburban infrastructure based on AHP-entropy method, Advanced Materials Research919-921 (2014), 1451–1456.

Q.F.

, Cong

, Yang

H.T.

, et al., Application of AHP based on entropy weight for risk identification in bridge construction, Advanced Materials Research243-249 (2011), 1871–1875.

Bakhoum

E.S.

and Brown

D.C.

, A hybrid approach using AHP–TOPSIS–entropy methods for sustainable ranking of structural materials, International Journal of Sustainable Engineering6(3) (2013), 212–224.

Walkenhorst

and Yasui

, Quantitative assessment of the benefits of trade facilitation, Oecd Papers2(3) (2009), 191–220.

Manuj

, Risk management in global sourcing: Comparing the business world and the academic world, Transportation Journal52(1) (2013), 80–107.

Xie

, Dong

, Hua

, et al., Safety evaluation of smart grid based on AHP-entropy method, Systems Engineering Procedia4(2012) (2012), 203–209.

10.

Manuj

, Risk management in global sourcing, Transportation Journal52(1) (2013), 80–107.

11.

Zhang

Y.R.

and Cao

W.F.

, Empirical research on logistics engineering based on AHP and entropy method, Applied Mechanics & Materials668-669 (2014), 1663–1666.

12.

, Wei

, Ni

X.L.

, et al., Evaluation of urban human settlement quality in Ningxia based on AHP and the entropy method, Ying Yong Sheng Tai Xue Bao25(9) (2014), 2700–2708.

13.

Barot

, Lourdeaux

, Burkhardt

J.M.

, et al., V3S: A virtual environment for risk-management training based on human-activity models, PRESENCE: Teleoperators and Virtual Environments22(1) (2013), 1–19.

14.

Arvai

and Post

, Risk management in a developing country context: Improving decisions about point-of-use water treatment among the rural poor in Africa, Risk Analysis32(1) (2012), 67.

15.

J.P.

, Li

, Wei

C.F.

, et al., Potentialities evaluation of regional land consolidation based on AHP and entropy weight method, Transactions of the Chinese Society of Agricultural Engineering25(5) (2009), 202–209(8).