2015年
财会月刊(20期)
ACADEMIC FRONTIERS
Research on Innovative Patterns of Financing of

作  者
Min Jian,Zhang Youtang

作者单位
(School of Management,Wuhan University of Technology,Wu Han,P.R. China,430070)

摘  要

Abstract: From the perspective of internal and external environment analysis, we construct the risk identification index system for overseas investment enterprises. Combined with the theory of comprehensive evaluation and risk early warning, the risk location system of overseas investment is established. The risk intelligence decision model is constructed by rough set theory, and the risk identification, risk location and risk decision of overseas investment are studied, and are empirically analyzed with cases in overseas investment.
Keywords: overseas investment; risk early warning; positioning system; rough set

1. Introduction
With the development of global economic integration, cooperation between enterprises in various countries is getting deeper everyday. And the advent of the fifth wave of mergers and acquisitions has set up the dominant position of the multi-national corporation. Meanwhile, the outbreak of the financial crisis also has made some developed countries become debt-ridden, and some foreign enterprises shrink in assets. This situation objectively provides a good opportunity for Chinese enterprises which are growing in strength and scale to enter the international market. However, many unfavorable factors, such as the international exchange rate fluctuations, the global economic downturn, the host country"s political barriers, cultural differences, financial risk, and labor conflicts have always been affecting and frustrating the overseas investment of Chinese enterprises. Faced with the severe and complicated situation at home and abroad, Chinese enterprises need to focus on the adjustment, closely track the situation changes, and overcome the impact of unfavorable factors, such as deepened trade barriers, increased costs and intensified risks in the implementation of the overseas investment. Thus, it becomes rather important for Chinese companies to accurately identify the risk of overseas investment, to scientifically explore the rules of risk evolution and to seek the rules for overseas investment risk decision.
Different from other project risks, risk of overseas investment companies is difficult to assess accurately and intuitively, mainly because that in overseas investment, it is difficult to unify the evaluation criteria and to obtain the accurate data, besides the impact of too many risk factors. In addition, guidance for enterprise investment behaviors in the monitoring process of the risk is not only a great emphasis for managers, but it can even directly help enterprise decision-makers to make judgment. But the traditional overseas investment risk decision-making model that only focus on deepening the theoretical research and methods does not succeed in finding the fuse which triggers the risks, nor does it succeed in further combining the theoretical researches with practical problems. Therefore, we should start from the perspective of enterprise managers to further study the risk decision-making model of overseas investment enterprises based on cases and data, and to lay a foundation for further analysis.
2. Risk Identification Index System for Overseas Investment of Enterprises
The enterprise overseas investment is faced with the influence of the internal and external environment changes, and the enterprise overseas investment risk is caused by internal and external environment changes. The external environments include the political and legal environment, the economic environment, the social and cultural environment, and the environment of the market; the internal environments include the financial environment, the market environment, the management environment, etc. Therefore, based on the analysis of the environment, we can construct the risk early warning index system for overseas investment.
As shown in figure 1, the risks of overseas investment can be divided into external risk indexes and internal risk indexes. The external risk indexes include political and legal risks, economic risks, socio-cultural risks and market risks, and the internal risk indexes include strategic decision-making risks, financial risks, trading risks and integrated risks. We can provide sources for overseas investment risk identification of enterprises through the identification of risk indexes, evaluation targets for the positioning of overseas investment through assessment of the risk indexes, and decision-making rules for the enterprises overseas investment through the screening of key risk indexes.
3. Risk Positioning System for Enterprise Overseas Investment
Based on the features of the enterprise overseas investment risk early warning index system, this paper constructs a risk location model for overseas investment enterprises which positions the overseas investment risks by adopting the comprehensive evaluation method as the basic evaluation method, together with the AHP, information entropy theory and the efficacy coefficient method. And the evaluation results are treated as the data sources for the risk decision model of enterprise overseas investment.
3.1 Weight Design of Risk IndexesAs we mentioned before, there are many risk factors affecting the overseas investment. The index system is complex, and many indexes lack the quantitative standards, and need to rely on the empirical judgments of the experts. Therefore, in this paper, the weight of qualitative index of overseas investment risk assessment is determined by AHP, and the weight of quantitative index of overseas investment risk assessment is based on the information entropy theory. The weights of the two together, determine the weight of overseas investment risk so that the real level of enterprise overseas investment risk is reflected to the best degree.
3.1.1 Determining the Weight of Qualitative Index
Considering the difficulty of qualitative index in quantitative standard, this paper adopts the Delphi method, together with the AHP to determine the weight of qualitative index. By issuing questionnaires and converting expert opinions into the identifiable evaluation results, the judgment matrixes are filled in, and the relative weights and the consistency indexes are calculated by expert judgment matrix. The judgment matrixes that do not meet the requirements are reconstructed or eliminated. Finally, the absolute weight of different indexes is calculated on the relative weight by using the matrix multiplication to determine the weight of the standard layer to the target layer.
3.1.2 Determining the Weight of Quantitative Index
By calculating the entropy, we can obtain the discrete degree of an index. The greater the discrete degree, the greater the impact of the index will have on the comprehensive evaluation, which means a greater need to explain the index and a greater weight of this index.
3.2 Evaluation of the Risk Indexes
3.2.1 Quantitative Index Evaluation Technology
Evaluation criteria of the index system in all quantitative indexes is acquired on the basis of the existing international standards and the international standard values of performance evaluation in the Appendix II of "Standard Values of Enterprise Performance Evaluation", and then adjusted according to different enterprises. We obtain the authentic data of the index value through surveys, and convert the quantitative index values into dimensionless values with the method of generalized function.
3.2.2 Qualitative Index Evaluation Technology
In order to achieve an objective and impartial evaluation of the qualitative indexes, this paper adopts the Delphi method to evaluate the qualitative indexes. Handing out the scoring tables of the qualitative indexes of the enterprises overseas investment risk identification indexes to the experts, and the scores of experts will be evaluated comprehensively. Finally, the final scores of different qualitative indexes will come out.
3.3 Positioning of the Risk Evaluation Results
3.3.1 Establishment of Risk Threshold Value
The risk threshold value is also called the risk critical value. According to the risk features of overseas investment, this paper divides the risk evaluation results into 4 grades: non warning, light warning, moderate warning and severe warning. The specific process is as follows:
Taking the overseas investment enterprises at the same period as samples, the actual value of the target enterprise index is xi, ximax stands for the maximum value of corresponding index of the sample enterprise, ximin stands for minimum value of corresponding index of the sample enterprise, xi stands for the average value of the sample enterprise, and ηi1, ηi2, ηi3 stand for the risk threshold value, that is, the critical value of the sample enterprise. After the dimensionless treatment of the index, the index is converted into the reverse risk index. When ηi3≤xi≤ximax, it means that the risk level of the index is severe warning; when ηi2≤xi<xi3, it means that the risk level of the index is moderate warning; when ηi1≤xi<xi2, it means that the risk level of the index is light warning; when timin≤xi≤ηi1, it means that the risk level of the index is non warning. As is shown in formula 1:
[ηi3=xi=i=0nxinηi2=xi-xi-ximin3ηi1=xi-2(xi-ximin)3]                     (1) 
3.3.2 Judgment of Risk Level
In order to facilitate the judgment of the risk level, we need to use the efficacy coefficient method to make the external environment risk index value of the enterprise overseas investment and the internal risk index value dimensionless, and define them in the range ofR∈[0,4]. We should assume as the actual risk value, tmax as the maximum value, tmin as the minimum value and η1, η2, η3 respectively as 3 critical values. Thus, we can give definitions as follows:
Definition 1: If η3≤ti≤tmax, then R∈[3,4], meaning that the risk level is severe warning and the assignment is 3.
Definition 2: If η2≤ti≤η3 , then R∈[2,3), meaning that the risk level is moderate warning and the assignment is 2. Definition 3: If η1≤ti<η2, then R∈[1,2), meaning that the risk level is light warning and the assignment is 1.
Definition 4: If tmin≤ti≤η1, then R∈[0,1) , meaning that the risk level is non warning and the assignment is 0.
After the R value is obtained, the risk level can be converted into the risk attribute value for the risk intelligence decision model of the enterprise overseas investment risk, for example 0, 1, 2, 3. In this paper, we mark the risk condition attributes as 0, 1, 2, 3, which respectively stands for non warning, light warning, moderate warning and severe warning; mark the decision attributes as 0, 1, 2, which respectively means "no risk or low risk, the next move can be conducted", "certain degree of risk, risk monitoring is needed in conducting next move" and "high risk, giving up on the next move is suggested".
4. Innovation in Risk Positioning of Enterprise Overseas Investment Based on the Intelligent Decision Model
4.1 Index Reduction of Enterprise Overseas Investment Risk
Generally speaking, risk attributes in the risk decision information table are not necessarily related to decision results. Attribute reduction is to delete some redundant and uncorrelated risk attributes while retaining the attributes associated with the decision results, and regarding them as a basis for decision-making.
Definition 5: Suppose S=(A,B) is the information chart, B⊆A and a∈B.
(1) If IB=IB-{a}, then risk a is redundant in B; otherwise, a is necessary in B;
(2) If all the attributes of B are necessary, then set B is independent;
(3) Suppose B′⊆B, if B′ is independent, and I′B=IB, then B′ is a reduction of B. Attribute set B may have several reduction s, and the set which consists of all the reductions of B is Red(B).
Definition 6: The intersection of all the reductions in attribute set B is the core of B, that is:
core(B)=⋂Red(B) (2) 
Because the core is the intersection of all the reductions, involved in each reduction of B, it is the most important attribute subset of B. Any element can not be removed without changing the classification capability.
This paper uses the genetic reduction algorithm which is based on attribute dependent degree to reduce the enterprise overseas investment risk index, namely use relative dependence of the heuristic information decision attributes on the condition attributes to guide the process of the next generation search operations. We calculate the relative core with the heuristic information, and add the relative core to the initial population of genetic algorithm GA to accelerate the convergence, and thus screening out the key risk indexes.
Rosetta, having the features of fast operation, detailed helping documents and user applicability, is chosen in this paper to implement the reduction of the risk monitoring indexes of enterprise overseas investment.
4.2 Excavation of Risk Decision Rules for Overseas Investment of Enterprises
Data updating consisting of decision rules in overseas investment projects is necessary in the excavation of risk rules. Because overseas investment projects are occurring all the time, people"s understanding of their risks is constantly changing. We can overcome the disadvantages of static decision making in the current complex and changeable environment by establishing a database that can dynamically update the number of rules.
Therefore, we can establish the general pattern of reasoning based on decision rules. Suppose there are decision rules as follows:
[f11∧f12∧…∧f1m→D=d1f21∧f22∧…∧f2m→D=d2…………………………fn1∧fn2∧…∧fnm→D=dn]           (3) 
Figure out its decision rule d* on the condition that c1∧c2∧…∧cm.
When combining the general pattern of decision rules based on decision rules and the risk decision model of overseas investment based on rough set, we can get the result S=(U,C⋃D,V,f). If the value range of risk conditional attribute a1 is {f11,f21,…,fn1}, the value range of risk conditional attribute a2 is {f12,f22,…,fn2}, and by such analogy, the value range of risk conditional attribute am is {f1m,f2m,…,fnm}; the value range of decision attribute D is{d1,d2,…,dn}, and di≠dj,i≠j,1≤i≤n,1≤j≤n,  f(1,d)=d1,f(2,d)=d2, by such analogy, f(n,d)=dn.
Suppose that C={a1,a2,…,am};
A=C⋃D,U={X1,X2,…Xn};f:U×A→V;
f(k,ai)=fki,1≤k≤n,1≤i≤m;
f(n+1,k)=ck,1≤k≤m;                   (4) 
Then S=(U,C⋃D,V,f) is an incomplete risk information system, and its decision rule inference problem evolves into dividing the discourse domain U into n types according to decision attribute D, and determines the decision attribute value d* corresponding to the n+1 decision target in the risk decision information chart.
4.3 Risk Intelligence Decision-making for Overseas Investment of Enterprises
Definition 7: If the risk attribute value of the enterprise c1∧c2∧…∧cm is consistent with one rule in the existing decision rule fi1∧fi2∧…∧fim→D=di, namely, if there exist the situation when c1=fi1,c2=fi2,…,cm=fim, then it can be said that the risk attribute value of the enterprise c1∧c2∧…∧cm activates one rule in the existing decision rule, and the decision attribute value d* of the decision object is directly derived from di.
When the risk attribute value of the enterprise  c1∧c2∧…∧cm fails to activate any rule in the existing decision rule, the decision attribute value d* is determined by risk condition attribute {a1,a2,…,am}.
If the decision situation of each risk condition attribute on dk,1≤k≤n is:
 [Card(i∈[dk]/f(i,a1)=c1)Card(dk)Card(i∈[dk]/f(i,a2)=c2)Card(dk)                                                 …Card(i∈[dk]/f(i,am)=cm)Card(dk)]   (5)  
It represents the support rate of risk condition attribute {a1,a2,…,am} to the evaluative risk attribute value c1∧c2∧…∧cm→d*=dk.
In the process of making decision rule inference, the importance of every risk attribute should be considered, so a judgment matrix is needed.
Definition 8: In the risk decision model of overseas investment S=(U,C⋃D,V,f), the judgment matrix of the risk attribute of the enterprise c1∧c2∧…∧cm is a n×m matrix:

 

 

 

         (6) 
Definition 9: In the risk decision model of overseas investment S=(U,C⋃D,V,f), the judgment vector of the risk attribute of the enterprise c1∧c2∧…∧cm is:
E=AI0   (7) 
Ais the normalized matrix of judgment matrix B;Vector I0 is the normalized vector of vector I, I=(Ia1,Ia2,…,Iam)T,Iai=1-Card(posC-ai(D))/Card(U),1≤i≤m,I is a vector made up by the normalization of different attributes:
[I=1-Card(posC-a1(D))Card(U)1-Card(posC-a2(D))Card(U)                  ⋯1-Card(posC-am(D))Card(U)]                        (8) 
If judgment vector, then one of its elements ei,1≤i≤n is the probability of emergence of ci in {f(j,a1)/f(j,a1)∈Va1 j∈[di]}, the probability of emergence of c2 in {f(j,a2)/f(j,a2)∈Va2 j∈[di]}, and by such analogy, the probability of emergence of cm in {f(j,am)/f(j,am)∈Vam j∈[di]}.
Judgment vector E=(e1,e2,…,en)T reflects the probability of decision value of the decision attribute D when the value of the risk condition attribute {a1,a2,…,am} is respectively{c1,c2,…,cm}. Because the decision attribute D is divided into {d1,d2,…,dn}, d*=dk can be evolved from the attribute value c1∧c2∧…∧cm, and k is decided by Max(e1,e2,…,en).
5. Empirical Analysis
5.1 Sample Selection and Data Acquisition
In order to ensure the authenticity, accuracy and science of the model simulation and evaluation results, this paper, according to the classification of fields, selects Chinese major large overseas investment projects in recent years as a case base. Except for parts of missing data and abnormal data samples, we chose 19 overseas investment enterprise as a decision rule mining research target and chose the case of Lenovo"s mergers and acquisitions of PC business of IBM as enterprises overseas investment risk decision-making research target.
In the data acquisition, this paper mainly selected the annual financial report data of the Shanghai and Shenzhen, including "The Consolidated Balance Sheet", "The Consolidated Income Statement", "The Consolidated Statement of Cash Flow" and the relevant non-financial data disclosed. These data is mainly acquired from Guo Tai"an Database, China Securities Regulatory Commission website, and various steel manufacturing corporation websites. For some macroeconomic indicators, this paper mainly acquires macroscopic data, especially the macroeconomic index data from "the IMF WEO Database", "the IMF IFS Database", "the World Bank WDI Database" and other databases. And the qualitative index is acquired through questionnaires and experts scoring tables. Due to the lag of the disclosure of the report of the listing corporation, the financial data of the sample is selected annually from the annual of its overseas investment. In order to establish risk decision-making information table, this paper selects TCL and other 10 enterprises as sample cases of enterprise overseas investment, as is shown in table 1.

 

 

 

 

 

 

 

 

 


This paper analyzes the case of the acquisition of the PC business of IBM Company by Lenovo Group as the research target for the risk decision of overseas investment. Based on the data information of investment behavior, the risk decision information is constructed. We use Rosetta software to carry out risk index reduction, and screen the key indicators that affect the investment behavior. We use the risk intelligence decision-making model of enterprise overseas investment to excavate the risk dynamic rules of the enterprise overseas investment, and the investment behavior is guided by the risk intelligence decision. To simplify the operation process, this paper selects the operational phase data of the investment behavior as the object of empirical research.
5.2 Determination of the Risk Positioning Results
After calculation, the risk positioning results of Lenovo in the operation stage of overseas investment are obtained, as is shown in table 2.
According to the above evaluation criteria and overseas investment risk index value, we can draw the final conclusion: The biggest risk that Lenovo faces in the decision-making stage comes from the external social cultural risk and the transaction risk in the enterprise. According to the actual case, the comprehensive assessment of the decision stage is 1, that is, “the risk monitoring is needed in the implementing the behavior of the next phase”.Similarly, conducting evaluation on the risk in the operational phase in other cases, we can get the risk index and the early warning and monitoring results of case U1 to U11,as is shown in table 3.
0 stands for “no risk or low risk, the next move can be conducted”; 1 stands for “certain degree of risk, risk monitoring is needed in conducting next move” and 2 stands for “high risk, giving up on the next move is suggested”.
5.3 Risk Index Reduction and Rules Excavation
According to the definition of 5 and 6, we can get the risk attribute reduction results calculated by Rosetta software. As is shown in table 4:

 

 

 

 

The result indicates that, in the operational phase of overseas investment, political and legal risk X1 , social and cultural risk X3 , financial risk X6 , and transaction risk X7 play a crucial role in deciding whether the overseas investment project can move on from the operational stage into the stage of integration and in deciding the degree of the risk.
According to formula 3 and 4, we can get the decision rules with Rosetta software, as is shown in table 5:
There are 6 rules in the operation stage of overseas investment:
(1) Rule f1: when financial risk=0, transaction risk=0, social and cultural risk=0, and political and legal risk=0, comprehensive assessment=0, meaning "no risk or low risk, the next move can be conducted".
(2) Rule f2: when financial risk=0, transaction risk=0, social and cultural risk=0, and political and legal risk=1, comprehensive assessment=0, meaning "no risk or low risk, the next move can be conducted".
(3) Rule f3: when financial risk=1, transaction risk=0, social and cultural risk=0, and political and legal risk=1, comprehensive assessment=0, meaning "certain degree of risk, risk monitoring is needed in conducting next move".
(4) Rule f4: when financial risk=1, transaction risk=1, social and cultural risk=1, and political and legal risk=1, comprehensive assessment=1, meaning "certain degree of risk, risk monitoring is needed in conducting next move".
(5) Rule f5: when financial risk=1, transaction risk=2, social and cultural risk=2, and political and legal risk=2, comprehensive assessment=2, meaning "high risk, giving up on the next move is suggested".
 (6) Rule f6: when financial risk=2, transaction risk=1, social and cultural risk=1, and political and legal risk=1, comprehensive assessment=2, meaning "high risk, giving up on the next move is suggested".
5.4 Risk Intelligent Decision
Through information processing of the data on the merger and acquisition of IBM PC business by Lenovo Group, we can initially construct the risk condition attribute values in the operation stage of the investment behavior; with the combination of risk decision rules, we can construct the intelligent risk decision table in the operation stage. As is shown in table 6:
Because the evaluated risk attribute values of the enterprise financial risk=1, transaction risk=1, social and cultural risk=1, and political and legal risk=0, not in correspondence with any existing rule fi1∧fi2∧…∧fim→D=di, the evaluated attribute value of the enterprise c1∧c2∧…∧cm can not activate any rule in existing decision rules, the decision attribute value d* is determined by risk condition attribute {a1,a2,…,am}.
According to formula 6, we can get the result:
B=[  0      0       0      0.5  1    0.5   0.5      00.5  0.5    0.5      0]
Normalized treatment:
A=[    0         0       0        10.667    0.5   0.5      00.333    0.5   0.5      0]
Normalized according to formula 8:
I0=[0.3750.187 50.187 50.25]
According to formula 7:
E=AI0=[    0         0       0        10.667    0.5   0.5      00.333    0.5   0.5      0][0.3750.187 50.187 50.25]=[0.2500.4380.312]
Because e2=max(0.250,0.438,0.312), namely the probability of d*=1 is 0.438, which is the highest in all the decision attributes, so the enterprise chooses "certain degree of risk, risk monitoring is needed in conducting next move".
6. Conclusion
This paper deeply studies the risk of overseas investment of enterprises, and establishes the risk location model and intelligent decision model of overseas investment based on the establishment of risk identification index system. Rosetta software is used to verify the accuracy of the research results. The main conclusions are as follows.
(1) In the process of selecting the key indexes for overseas investment risk intelligent decision-making, the dynamic rule excavation and decision making of investment risk, rough set and genetic algorithm are adopted. Through empirical analysis, the following conclusions are drawn: under the incomplete information discrimination, the overseas investment risk intelligent decision model will still be able to obtain the corresponding evaluation and decision-making effectiveness, and can guide enterprises overseas investment decision in a certain extent.
(2) In the operation of the intelligent decision-making model for overseas investment risk, the Rosetta software and Excel computing tools are used to explore how rough set can be used to achieve the implementation platform of large sample computation. The results of empirical analysis show that: in the computation of the index operation and the knowledge inference of the risk intelligence decision model, Rosetta and Excel can effectively simplify the operation process and enhance the practicality of the system, and it will help in the application in a wider range.
References
Dong Lijun. Empirical Test of Overseas Investment Performance of Listing Corporation[J]. Financial Monthly, 2012(2).
Zeng Daorong, Tan Mingjun. Effective Ways of Improving the Financial Vitality of Enterprise during the Financial Crisis[J]. Management World, 2010(10).
Zhang Youtang, Xiao Hui. Regional Analysis of Transnational Investment Risk in Mining Enterprises Based on HHM and RFRM[J]. Financial Monthly, 2011(21).
Axson D J.Scenario Planning:Navigating Through Today"s Uncertain World[J].Journal of Accountancy,2011(3).
Jan Bazan,Marcin Szczuka,Jakub Wroblewski.A New Version of Rough Set Exploration System [J].Lecture Notes in Artificial Intelligence,2010(3).