1. bookAHEAD OF PRINT
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Otwarty dostęp

Enterprise Financial Risk Early Warning System Based on Structural Equation Model

Data publikacji: 15 Jul 2022
Tom & Zeszyt: AHEAD OF PRINT
Zakres stron: -
Otrzymano: 11 Feb 2022
Przyjęty: 13 Apr 2022
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Introduction

With the globalization of the world economy, companies have found many opportunities in it. But there are many crises hidden in this. If not discovered in time, these crises will evolve into financial crises and eventually lead to corporate bankruptcy. As a tool for predicting financial risks, the early-warning system can judge the current development status of the enterprise based on financial indicators. It can discover and solve the enterprise's management loopholes and operational defects in time and reduce the probability of financial crises. And cash flow is the core of business operation and financial status, and it is also the foundation for the survival and development of the business [1]. Its values in various activities such as operation, investment, and fund-raising can reflect the pros and cons of the business. The financial risk early-warning indicators based on cash flow are the most valuable in evaluating the value of enterprises. Effective cash flow indicators to build a financial risk early warning system to ensure that the company can carry out daily operations normally and efficiently is of great significance to corporate managers.

Research Design
Index selection

The financial risk early-warning indicators selected in this paper are mainly cash flow indicators. This article divides the financial status (W) into four dimensions: solvency (Z1), cash-capability (Z2), profitability (Z3), and sustainable development capability (Z4).

Among them, solvency is divided into the cash debt protection ratio (X1) and cash current debt ratio (X2). The cash-earning capacity is divided into the sales cash ratio (X3), the net cash flow from operating activities per share (X4), and the total asset-to-cash ratio (X5). Profitability is divided into net profit cash content (X6) and earnings per share (X7). Sustainability is divided into operating income growth rate (X8), net assets growth rate (X9), and shareholder equity growth rate (X10).

Sample company design

This article selects all A-share ST companies in the industry as of May 31, 2019, and matches each ST company with a financially sound company [2]. The non-ST companies that choose to match mainly follow the following principles. The first financial crisis company and the financially normal company should be in the same or similar industry, and the asset scale of the two should be close. If normal companies with similar asset sizes cannot be found in the same or similar industries, companies in other industries with the same risks can be considered. Second, the study period is the same. That is, the research time is from 2016 to 2018. Third, the ratio of 1:1 is adopted. The sample size of ST companies and non-ST companies should be equal.

In selecting data, this paper selects the data of the three years from 2016 to 2018 for analysis. This is because the company's financial difficulties did not happen suddenly but went through a continuous and long process. The consecutive losses in the last two years are the main reason for the company being ST. Although ST Company maintained a numerically good level in its annual report for t-3 years, it continued to lose money in the following years. It can be seen that the difference between ST companies and non-ST companies should be reflected int-3 years (that is, 2016 in this article). Therefore, this paper selects relevant indicators for t-3 years and the following two years for financial risk warning.

After screening and deleting companies with incomplete financial data, a total of 59 companies remain [3]. We paired them with non-ST companies, a total of 118 companies. Among them, the top 49 companies are selected as the research samples, and the remaining 10 companies are selected as the test samples. The data in this article mainly comes from Hexun.com, Juchao.com, and Guotai Junan Database.

Structural equation model

A structural equation model is a method to establish a causal relationship between variables, including easily available observed variables and difficult to measure latent variables. In today's various research fields, most exploratory factor analyses, such as factor analysis and principal component analysis, are used for the statistical processing of data. Still, this type of analysis has certain limitations [4]. The first type of analysis assumes that all factors will affect the measurement items. Still, there may be situations where some factors do not affect the measurement items in actual research. The second type of analysis assumes that the residuals of the measure items are independent of each other. However, in actual research, there is more or less a certain correlation between the residuals. The structural equation model can exclude the influence of residuals on the results. The third type of analysis stipulates that all factors are independent, but the variables should be related. The structural equation model can handle the relationship between multiple causes and results. Therefore, this article uses a structural equation model to explore the relationship between variables. The structural equation model consists of the following two models. One is the measurement model. It reflects the relationship between latent variables and measured variables. Y=λYη+ε Y = {\lambda _Y}\eta + \varepsilon X=λXξ+σ X = {\lambda _X}\xi + \sigma

Y and X are the endogenous and exogenous observed variables, respectively. η and ξ are the endogenous and exogenous latent variables, respectively. λY and λX are coefficients. It reflects the relationship between observed variables and latent variables. ɛ and σ are the residual matrix.

The second is the structural model. It reflects the relationship between latent variables. η=Bη+γξ+ζ \eta = B\eta + \gamma \xi + \zeta

B and γ are coefficient matrices. It respectively reflects the relationship between the constituent factors of endogenous and exogenous latent variables. ζ is the residual matrix.

Model fitting

This article standardizes the data. Unify all data into dimensionless data. Secondly, the Cronbach's Alpha coefficient is calculated to be 0.657. It is within the acceptable range. This shows that the credibility of the sample data is better [5]. Finally, use Amos24.0 software to calculate the structural equation model. We use maximum likelihood estimation to fit the model and build the model, as shown in Figure 1.

Figure 1

Diagram of financial risk influencing factors

Validity analysis

Validity analysis reflects the potential characteristics of indicator variables to be measured. After calculating the chi-square value is 40.938, the degree of freedom is 31, and the chi-square/degree of freedom = 1.321. The main fitting indexes of the comprehensive model are shown in Table 1 to Table 3. The values of the absolute fitting index, relative fitting index, and simple fitting index are all within the standard range. Most of them are close to the theoretically expected value [6]. This shows that the model has a high degree of fit and a good overall fit.

Absolute Fit Index

index Numerical value
Absolute Fit Index γ2/df 1.321
RMSEA 0.033
RMR 0.043
GFI 0.973
AGFI 0.952
NCP 0.938

Relative Fitting Index

index Numerical value
Relative fit index NFI 0.984
RFI 0.977
IFI 0.996
TLI 0.994
CFI 0.996

Reduced Fit Index

index Numerical value
Parsimonious fit index AIC 88.938
CAIC 201.344
PNFI 0.678
PGFI 0.549
PCFI 0.686
BIC 177.344
Reliability analysis

Reliability analysis uses two indicators, combined reliability (CR) and average variance extraction (AVE), to describe. The combined reliability (CR) reflects the inherent similarity between the observed variables to which the latent variable belongs. The greater the degree of similarity, the higher the consistency of the observed indicators. It is generally believed that the combined reliability of latent variables should be greater than 0.6. The extracted average variance value (AVE) reflects the internal consistency of structural variables, and the AVE value should generally be greater than 0.5. It can be obtained by calculation that Z3 does not reach the standard level [7]. This is because it is difficult to obtain a model with all indicators meeting the requirements in practice. However, from the perspective of other latent variables, P, Z1, Z2, Z4 combined reliability, and average variance extraction have reached the standard level. It can be seen that the model has good credibility.

Evaluation results

After the structural equation model is fitted, the weight value between the observed and latent variables can be determined according to the output result. The expressions of the latent variables obtained are as follows: Z1=0.966*X1+0.961*X2Z2=0.643*X3+0.721*X4+1.056X5Z3=0.275*X6+0.85*X7Z4=0.841*X8+0.997*X9+0.989X10 \matrix{ {{Z_1}} \hfill & = \hfill & {0.966*\,{X_1} + 0.961*{X_2}} \hfill \cr {{Z_2}} \hfill & = \hfill & {0.643*\,{X_3} + 0.721*\,{X_4} + 1.056{X_5}} \hfill \cr {{Z_3}} \hfill & = \hfill & {0.275*\,{X_6} + 0.85*\,{X_7}} \hfill \cr {{Z_4}} \hfill & = \hfill & {0.841*\,{X_8} + 0.997*\,{X_9} + 0.989\,{X_{10}}} \hfill \cr }

Under the processing of the structural equation model, the relationship between the various indicators under the latent variables can be obtained. Still, the difference in the indicator values between the ST company and the normal company cannot be analyzed. Therefore, logistic regression analysis is performed on the latent variables processed and calculated by the structural equation model to determine the probability of each company falling into a financial crisis [8]. This establishes a financial risk early-warning model.

Logistic regression analysis

Logistic regression analysis transforms the problem of predicting whether the company will fall into financial distress or not into calculating the probability of the enterprise falling into a financial crisis. Suppose that the probability of a company falling into financial distress is P, and the value of P ranges from 0 to 1. When P is greater than a certain set value, the company will fall into a financial crisis. Its function is as follows: Logit(P)=ln(P1P)=a0+a1Z1+a2Z2++anZn Logit\left( P \right) = \ln \left( {{P \over {1 - P}}} \right) = {a_0} + {a_1}{Z_1} + {a_2}{Z_2} + \ldots + {a_n}{Z_n}

Simplification can get: P=e(a0+a1Z1+a2Z2++anZn)1+e(a0+a1Z1+a2Z2++anZn) P = {{{e^{\left( {{a_0} + {a_1}{Z_1} + {a_2}{Z_2} + \ldots\, + {a_n}{Z_n}} \right)}}} \over {1 + {e^{\left( {{a_0} + {a_1}{Z_1} + {a_2}{Z_2} + \ldots\, + {a_n}{Z_n}} \right)}}}} Where X1, X2Xn is the independent variable. a0 is a constant term. a1, a2an is the regression coefficient.

Use SPSS software to perform Logistic regression. For a company in financial distress, P is set to 1. Otherwise, it is set to 0. According to the obtained equation, 0.5 is chosen as the critical value of the financial crisis. If the value of P>0.5, the sample is regarded as a company that has fallen into a financial crisis [9]. If the P-value is less than 0.5, it is considered that the company is operating well and does not have excessive financial risks. Logistic regression is performed on the four latent variable indicators calculated above, and the probability value calculation equation of falling into a financial crisis can be obtained as follows: P=e0.0520.02*Z1+0.016*Z21.122*Z30.448Z41+e0.0520.02*Z1+0.016*Z21.122*Z30.448Z4 P = {{{e^{ - 0.052 - 0.02*{Z_1} + 0.016*{Z_2} - 1.122*{Z_3} - 0.448\,{Z_4}}}} \over {1 + {e^{ - 0.052 - 0.02*{Z_1} + 0.016*{Z_2} - 1.122*{Z_3} - 0.448\,{Z_4}}}}}

Model-checking

According to the equations obtained above, a financial risk early warning model is established, and 10 sets of test samples are used for testing. Judging from the effect of model prediction, 22 of the 30 samples were tested by the model in the discrimination of crisis enterprises, and the model's prediction accuracy rate was 73.33%. This shows that the model has a strong ability to test financial crises [10]. The 8 unchecked samples are from 2016 and 2017. The probability value of 5 samples is at the level of 0.4 to 0.5, which is close to the critical value of the financial crisis. And as time goes by, the probability value increases year by year. In addition, in the judgment of normal enterprises, only 3 of 30 samples were judged to be in financial crisis, and the misjudgment rate was 10%. The financial crisis occurred in 2016 and 2017. This shows that the model can make a more accurate judgment on the company's actual situation through financial data and encourage corporate managers to use various strategic decisions to resist corporate risks. In the end, all enterprises will reach normal levels in 2015.

Design of financial early warning system
Setting of corresponding indicators for early financial warning

Companies and their specific business projects, activity structure, etc., set financial indicators to use the above factors as the basis for measuring business results. However, the indicators set by most companies have the characteristics of coincidence [11]. Common indicators include cash flow, debt ratio, net income, etc. Specifically, the enterprise managers can determine the appropriate positioning and indicators based on their specific development situation.

Choose an early warning model

When constructing an early warning model for an enterprise, an in-depth analysis of the model can be carried out based on the current research results according to the corresponding category. The research results show that most computers with fewer independent variables will produce corresponding errors inaccuracy. The main purpose of an enterprise set up a financial early warning system is to ensure the smooth operation of the enterprise [12]. The principal component analysis method can be used for research. This research method needs first to determine the main components and set the weights for the corresponding components. The most popular form of a complete early warning model. The specific model building steps can be the following three points:

(1) Calculate the relevant coefficients and create a relatively stable variable coefficient index. The specific matrix form can be as follows: W=(ω11ω12ω1qω21ω22ω2qSSSωq1ωq2ωqq) W = \left( {\matrix{ {{\omega _{11}}} & {{\omega _{12}}} & \cdots & {{\omega _{1q}}} \cr {{\omega _{21}}} & {{\omega _{22}}} & \cdots & {{\omega _{2q}}} \cr S & S & \cdots & S \cr {{\omega _{q1}}} & {{\omega _{q2}}} & \cdots & {{\omega _{qq}}} \cr } } \right)

Specifically, the following formula can be used for calculation: ωij=l=1x(NliN¯i)(NljN¯j)l=1x(NliN¯i)2l=1x(NljN¯j)2 {\omega _{ij}} = {{\sum\limits_{l = 1}^x {\left( {{N_{li}} - {{\bar N}_i}} \right)\left( {{N_{lj}} - {{\bar N}_j}} \right)} } \over {\sqrt {\sum\limits_{l = 1}^x {{{\left( {{N_{li}} - {{\bar N}_i}} \right)}^2}\sum\limits_{l = 1}^x {{{\left( {{N_{lj}} - {{\bar N}_j}} \right)}^2}} } } }}

(2) Simplify the setting of the characteristic equation first. The purpose is to export the vector of each eigenvalue f1 (i = 1, 2, ⋯, q).

Basic structure of enterprise financial early warning system
Financial risk information subsystem

Most innovative enterprises have reached a current operating level. They have built a network information system with relatively complete functions. The information acquisition module can play a very important role in the entire early warning system. The financial department provides smoother information to enhance the stability of enterprise development, and the financial department's effectiveness can directly improve the accuracy of actions. Therefore, financial personnel needs to effectively use risk information systems to scientifically respond to market pressures and conduct real-time monitoring of brain drain and investment ratios. At the same time, it can also show the potential risks of the enterprise in time.

Financial risk index calculation subsystem

The system can scientifically weigh the effectiveness of cash inflows and the rationality of increased sales data. The financial early warning system of an enterprise can mainly collect and verify risks. The technology, as mentioned earlier, can check the validity based on a large amount of data to promote the overall improvement of the quality of the subsystem.

Financial risk comprehensive assessment subsystem

The evaluation of the function of the subsystem is mainly to measure the enterprise's financial situation comprehensively. The results obtained can be divided into positive and negative. If the evaluation result is negative, relevant information needs to be collected from the background. This provides companies with the necessary information report content and reduces the system's misjudgment.

Financial risk warning subsystem

This system can play a role based on the establishment of the system, as mentioned above. Obtain various target information from the above system. If the risk has reached the corresponding warning level, the system can directly provide warning information to the manager or other relevant personnel. Only designated personnel can enter the system to cancel the alarm. The system can comprehensively improve the professional level of the financial early warning system and ensure that it can perform the expected functions in each professional state.

Financial risk control subsystem

The main function of the system is to adjust and alert the corresponding project network ports. Technicians can directly use the system to verify the effectiveness of the decision. If the displayed information is valid, the established strategy can be directly applied. If it is ineffective, a professional team of financial management should be used to reduce risks. Figure 2 is an illustration of the structure of the financial early warning system.

Figure 2

Overview of the structure of the enterprise financial early warning system

Conclusion

This paper obtains a financial risk early-warning model with a good fitting effect through empirical research. It can provide early warning and prevention of financial risks in the business process and reduce the possibility of risks. This allows the company to maintain good operating conditions. Good operation is inseparable from the monitoring of financial status, especially in terms of cash flow. The notable sign of a company falling into a financial crisis is the shortage of cash and easily realizable assets, accompanied by a large amount of debt that needs to be paid off. And these signs will be reflected through the financial risk early-warning model.

Figure 1

Diagram of financial risk influencing factors
Diagram of financial risk influencing factors

Figure 2

Overview of the structure of the enterprise financial early warning system
Overview of the structure of the enterprise financial early warning system

Reduced Fit Index

index Numerical value
Parsimonious fit index AIC 88.938
CAIC 201.344
PNFI 0.678
PGFI 0.549
PCFI 0.686
BIC 177.344

Absolute Fit Index

index Numerical value
Absolute Fit Index γ2/df 1.321
RMSEA 0.033
RMR 0.043
GFI 0.973
AGFI 0.952
NCP 0.938

Relative Fitting Index

index Numerical value
Relative fit index NFI 0.984
RFI 0.977
IFI 0.996
TLI 0.994
CFI 0.996

Rai, K., Dua, S., & Yadav, M. Association of financial attitude, financial behaviour and financial knowledge towards financial literacy: A structural equation modeling approach. FIIB Business Review., 2019; 8(1):51–60 RaiK. DuaS. YadavM. Association of financial attitude, financial behaviour and financial knowledge towards financial literacy: A structural equation modeling approach FIIB Business Review 2019 8 1 51 60 10.1177/2319714519826651 Search in Google Scholar

Alhassany, H., & Faisal, F. Factors influencing the internet banking adoption decision in North Cyprus: an evidence from the partial least square approach of the structural equation modeling. Financial Innovation., 2018; 4(1):1–21 AlhassanyH. FaisalF. Factors influencing the internet banking adoption decision in North Cyprus: an evidence from the partial least square approach of the structural equation modeling Financial Innovation 2018 4 1 1 21 10.1186/s40854-018-0111-3 Search in Google Scholar

Nurius, P. S., Fleming, C. M., & Brindle, E. Life course pathways from adverse childhood experiences to adult physical health: A structural equation model. Journal of Aging and Health., 2019; 31(2):211–230 NuriusP. S. FlemingC. M. BrindleE. Life course pathways from adverse childhood experiences to adult physical health: A structural equation model Journal of Aging and Health 2019 31 2 211 230 10.1177/089826431772644828845729 Search in Google Scholar

Ahmad Sabir, S., Mohammad, H., & Kadir Shahar, H. The role of overconfidence and past investment experience in herding behaviour with a moderating effect of financial literacy: evidence from Pakistan stock exchange. Asian Economic and Financial Review., 2019; 9(4):480–490 Ahmad SabirS. MohammadH. Kadir ShaharH. The role of overconfidence and past investment experience in herding behaviour with a moderating effect of financial literacy: evidence from Pakistan stock exchange Asian Economic and Financial Review 2019 9 4 480 490 10.18488/journal.aefr.2019.94.480.490 Search in Google Scholar

Shapiro, S. L., Reams, L., & So, K. K. F. Is it worth the price? The role of perceived financial risk, identification, and perceived value in purchasing pay-per-view broadcasts of combat sports. Sport Management Review., 2019; 22(2):235–246 ShapiroS. L. ReamsL. SoK. K. F. Is it worth the price? The role of perceived financial risk, identification, and perceived value in purchasing pay-per-view broadcasts of combat sports Sport Management Review 2019 22 2 235 246 10.1016/j.smr.2018.03.002 Search in Google Scholar

Chatterjee, S., & Bhattacharjee, K. K. Adoption of artificial intelligence in higher education: A quantitative analysis using structural equation modelling. Education and Information Technologies., 2020; 25(5):3443–3463 ChatterjeeS. BhattacharjeeK. K. Adoption of artificial intelligence in higher education: A quantitative analysis using structural equation modelling Education and Information Technologies 2020 25 5 3443 3463 10.1007/s10639-020-10159-7 Search in Google Scholar

Foster, B., Johansyah, M. D., & City, B. The effect of product quality and price on buying interest with risk as intervening variables (study on Lazada.com site users). International Journal of Innovation, Creativity and Change., 2019; 9(12):66–78 FosterB. JohansyahM. D. CityB. The effect of product quality and price on buying interest with risk as intervening variables (study on Lazada.com site users) International Journal of Innovation, Creativity and Change 2019 9 12 66 78 Search in Google Scholar

Wijethilake, C., & Lama, T. Sustainability core values and sustainability risk management: Moderating effects of top management commitment and stakeholder pressure. Business Strategy and the Environment., 2019; 28(1):143–154 WijethilakeC. LamaT. Sustainability core values and sustainability risk management: Moderating effects of top management commitment and stakeholder pressure Business Strategy and the Environment 2019 28 1 143 154 10.1002/bse.2245 Search in Google Scholar

Jiao, Y. & Zheng, Q. Urea Injection and Uniformity of Ammonia Distribution in SCR System of Diesel Engine. Applied Mathematics and Nonlinear Sciences., 2020; 5(2): 129–142 JiaoY. ZhengQ. Urea Injection and Uniformity of Ammonia Distribution in SCR System of Diesel Engine Applied Mathematics and Nonlinear Sciences 2020 5 2 129 142 10.2478/amns.2020.2.00004 Search in Google Scholar

Wang, Ying and Chen, Yanfei. “Evaluation Method of Traffic Safety Maintenance of High-Grade Highway” Applied Mathematics and Nonlinear Sciences., 2021; 6(1): 65–80 WangYing ChenYanfei “Evaluation Method of Traffic Safety Maintenance of High-Grade Highway” Applied Mathematics and Nonlinear Sciences 2021 6 1 65 80 10.2478/amns.2021.1.00007 Search in Google Scholar

Anwar, A., Thongpapanl, N., & Ashraf, A. R. Strategic imperatives of mobile commerce in developing countries: the influence of consumer innovativeness, ubiquity, perceived value, risk, and cost on usage. Journal of Strategic Marketing., 2021; 29(8):722–742 AnwarA. ThongpapanlN. AshrafA. R. Strategic imperatives of mobile commerce in developing countries: the influence of consumer innovativeness, ubiquity, perceived value, risk, and cost on usage Journal of Strategic Marketing 2021 29 8 722 742 10.1080/0965254X.2020.1786847 Search in Google Scholar

Polecane artykuły z Trend MD

Zaplanuj zdalną konferencję ze Sciendo