The performance evaluation model of an empirical analysis for e-business based on data envelopment analysis

• i.    Research background

There is no doubt that the e-business has played the most important role in the process of the world economy globalization. The governments, enterprises and scholars have all shown great attentions on the electronic business for its final goals of strengthening their capabilities during the international competition. Taking all the other countries which have achieved better and faster development in this respect into consideration, China is still lagging behind in the construction of e-business, especially in the performance evaluation, which is the key point in what direction the e-business will go. During the past recent years, the performance evaluation concerned with e-business has not reached the expected achievements. Until now, the inherent system of the e-business performance evaluation is not clear and all the concerned studies are not linked each other. Obviously, this kind of situation can not guarantee the further study of the e-business. Actually, the e-business asks for a new set of performance evaluation system, which is not running the same way as the old one does.

This paper intends to focus its attention on such problems, that is, what factors influence e-business implementation and how to evaluate the hereof mentioned implementation effect. Based on the methods of data envelopment analysis whose excellent contribution to the performance evaluation has been proved by most aspects of the economics, we herein bring forwards a new model aiming at the provision of the performance evaluation model for e-business.

• II.    FACTORS ANALYSIS OF ELECTRONIC COMMERCE IMPLEMENTATION

The clear definition of successful implementation of e-business is the precondition for the performance evaluation. As far as the concerned studies, we have no other way but point out that the successful implementation of e-business is a multi-dimension concept. What makes us come to such a conclusion is that only one index or two can hardly describe the implementation of e-business and the judgment of its success depends on in what criterion the judger will choose.

Taking most of the concerned studies into consideration, we propose a modified success model, which includes system quality, information quality and the service quality. The model proposed here are outcomes of hundreds of concerned researches both from the theoretical and empirical point of view, occupying a prior authority and reliability. Besides the factors mentioned above, in this model we especially bring forward the satisfaction degree of the e-business system operation, which offers the clear and veracious direction of e-business system. However, this innovation has not been reached in the former studies. In the model, the information system is regarded as a whole process, including time and space dimension. The hereof system is firstly set up with a judgment of satisfaction for its operation. Therefore, the system quality and information quality both decide the performance of the model. There is no doubt that the effect can be positive or negative as well. Actually, critical success factors are the limited number of areas in which results, if they are satisfactory, will insure successful competitive performance for the organization. They are the few key areas where things must go right for the business to flourish. As a result, the critical success factors are areas of activity that should receive constant and careful attention from management.

• III.    PERFORMANCE EVALUATION MODEL BASED on Data Envelopment Analysis method

The data envelopment analysis (DEA) was firstly brought out by American operational researcher named A. Charnes and W. W. Cooper, which is widely used in the mathematics in evaluating the performance or efficiency. Now this part of the thesis will set up a model based on the DEA method.

In the following model, we regard the e-business to be targeted as AMR. The input factors for the performance evaluation are lined as a = (a ,a2,—, a )T, and the output factors for the production are lined as p = (д, д,—, д).

The number of the target e-business is q , AMR_} (1 <  j <  q ) . Each target e-business has its own input and output factors set, such ^as a j = ( a _v, a 2 j , - , a _pj ) ^T > 0 , Д = ( e _v, ^ 2 , , - , в ,) T > 0

and a, > 0; в > 0; i = 1,2,—, p; 5 = 1,2,—, r . All the proportions for the input and output factors are: X = (x„x2,...,xp)T, y = (y„y2,-,ys)T. xi is the proportion for the input factor of i and y is the proportion for the output factor of s . Each AMR has its own efficiency evaluation index k , and we can choose the proper proportion x and y to make k < 1.

, = ye j   xTa,

r

E y e

• 5 =1

pp

Ex,a i ij i =1

Now, we can evaluate the efficiency of target j . Normally, the higher k means higher efficiency of the target e-business, which makes the following method useful that is we max k can choose the proper proportion to check the highest k to match the best target. Taking this idea into the construction of our research, we set up the following model: r

E y e . о

_ 5 = 1

pr x,-a,-, i    ij0

r

E y e .

s . t .     ^-----< 1, j = 1,2, — , q

E xa i =1

x = ( x 1 ,x 2 , - , x p ) T > 0 , y = ⁽ y j , y 2 , ^- , y ^{) T} > ⁰

In the model, x > 0 means x, > 0 for every i = 1,2, —, p, and there will be some i0 (1 < i0 < p), which makes sure that x, > 0. Same result comes to y > 0.

Further more, we can take the Charnes-Cooper method to make the changes as follows:

_t _    ¹    ,   £ = tx , n = ty .

t     T x a0

And there will be the following linear programming:

max k . = n ^T P 0

( P ) ^

s . t .

Е^та . - n T в . >  0, j = 1,2, — . q

£ >  0 n >  0

If we change the above linear programming into dual programming, we can reach the following:

min 6

^^ X j « j + r ⁺ = 6a 0

j = 1

( D ) ^

jL Xjpj - r - = 6во j=1

Aj > 0,  j = 1,2, — , q r+ > 0,  r- > 0

• 6 is in free condition. r ⁺ is laxity variable and r ^- is residual variable. We can draw this conclusion that if the linear programming has best value, it must be in the condition of k * = 6 * <  1 . If the value has the result of j ₀

e >  0 , n >  0 , and k * = 1 , it means the best j ₀

cooperation.

• iv. The empirical analysis

• A.    The Performance Evaluation Indexes

Based on the above model, we herein choose four aspects of main variables to evaluate the e-business implementation, such as e-business system  quality  measures, e-business information  quality  measures, e-business service    quality   measures, e-business implementing effect. Under all the main variables, there are also some more detail variables, which can be reached in the following Tab. I.

Due to the characteristic of the DEA method, we should select one specific e-business as the target one for the further details research. Taking data availability into consideration, this paper chooses Shanghai as the target e-business and deploys the further study. Based on the DEA method, we need other regions’ data to evaluate the hereof target’s performance evaluation. So, we also gather 30 other e-business targets’ data, which will be embedded in the model.

Taking all the variables into the model proposed above, we collect all the data from the 30 e-business targets, which are all made dimensionless by the concerned soft. After the calculation, the Shanghai’s performance evaluation details can be showed in Tab. II from the year 2002 to 2010.

From Tab. II, the performance degrees have been on the way upwards in the year from 2002 to 2005, which means Shanghai has the harmonious e-business performance. And in the year of 2006, its e-business performance had a little bit decrease. From the year of 2007, the e-business performance were still on the upwards way.

However, from the year of 2006, the performance of Shanghai’s e-business touched its peak level 0.76 and then turned down. By the year of 2010, its performance was 0.70, up to 7% off the max. This decrease means Shanghai e-business met its hard time while all the main variables such as e-business system quality measures, e-business information quality measures, e-business service quality measures, e-business implementing effect can not fit each other quite well.

table i.   THE VARIABLES SET FOR PERFORMANCE EVALUATION

Standard hierarchy	Rule hierarchy	Quota hierarchy
E-business system quality measures (X 1 )	X 11 ： Usability, Ease of Use	X 111 ： Download Time
		X 112 ： Help Features
		X 113 ： Intuitiveness
		X 114 ： Attractiveness
	X 12 ： ependability, Reliability	X 121 ： System Responsiveness, Response Time
	X 13 ： Usefulness, Functionality	X 131 ：    Versionability
		X 132 ： Transaction Capabilities
	X 14 ： Security	X 141 ： Scalability
		X 142 ： Interactivity
E-business information quality measures (X 2 )	X 21 ： Customer Preference Information	X 211 ： Understandability
		X 212 ： Completeness
		X 213 ： Customer Feedback Capability
		X 214 ： Transaction Capabilities
	X 22 ： Customer Preference Information	X 221 ： Understandability
		X 222 ： Completeness
		X 223 ： Customer Information Integration Across Multiple Channels
		X 224 ： Currency
	X 23 ： Competitive Intelligence	X 231 ： Dynamic Content
		X 232 ： Content Personalization
		X 233 ： Variety of Information
E-business service quality measures (X 3 )	X 31 ： Liabilities	X 311 ： Sending in Time
		X 312 ： Easy Understanding
		X 313 ： On line FAQ
		X 314 ： Integrity of Staff
E-business implementing effect (X 4 )	X 41 ： Reachable Effect	X 411 ： Growth in Customer Base
		X 412 ： Increased Sales
		X 413 ： Market Share
		X 414 ： Return on Investment
		X 415 ： Customer lock-in
		X 416 ： Sales Process Efficiency

THE PERFORMANCE EVALUATION DEGREES

Year	2002	2003	2004	2005	2006	2007	2008	2009	2010
Performance Degree	0.50	0.54	0.57	0.61	0.58	0.65	0.76	0.74	0.70

All the data are achieved by the author with the software EMS1.3.

It is doubtless that the e-business performance can hardly always go the right way as expected. Therefore, the macro-control system is needed to correct the e-business while it goes the wrong way. The problem left now is how to find the right way to correct the e-business performance and what is the macro-control system.

• B.    Sensitivity Analysis of the Performance Degree

The sensitivity analysis is widely used in the economic analysis. Normally, the sensitivity analysis can be reached if two variables have the function connection. In the model proposed, sensitivity means the 1% change of the variables set can arouse the percentages of the performance degree in the model. Keeping all other e-business data stable, we then change the specific factor variables by 1% up, and then get new performance evaluation degrees for the concerned years.

The degrees achieved can be reached in the Tab. Ш table ii.   SENSITIVITY ANALYSIS FOR SPECIFIC FACTORS

Year	2002	2003	2004	2005	2006	2007	2008	2009	2010
Sensitivity	2.14	2.56	2.72	-0.17	3.17	3.88	-1.77	-2.13	-2.22
Former Degree	0.50	0.54	0.57	0.61	0.58	0.65	0.76	0.74	0.70
Modified Degree	0.51	0.57	0.59	0.61	0.60	0.68	0.74	0.72	0.68
All the data are achieved by the author with the software EMS1.3 on the modified data.

At the same time, we keep all the other data stable, and then change the hereof factor variables by 1% up to check the performance degree sensitivity level to the specific factors. All the outcome data are showed in the Tab. IV.

TABLE III.   SENSITIVITY ANALYSIS FOR BALANCED FACTORS

Year	2002	2003	2004	2005	2006	2007	2008	2009	2010
Matching Sensitivity	1.11	1.74	2.07	-1.08	1.17	1.68	1.49	2.53	3.82
Former Degree	0.50	0.54	0.57	0.61	0.58	0.65	0.76	0.74	0.70
Modified Degree	0.51	0.55	0.59	0.60	0.59	0.66	0.77	0.76	0.73

All the data are achieved by the author with the software EMS1.3 on the modified data.

As we known, e-business performance depends on the inter-cooperation of the concerned variables such as system quality measures, information quality measures, service quality measures and implementing effect. Only the main variables mentioned above fit each other quite well, e-business can have the excellent performance. From what we have achieved in Tab. Ш and Tab. V, it is easy to draw the conclusion that Shanghai e-business performance is sensitive to the specific factors variables of service quality measures, implementing effect, which means the performance degree of Shanghai can be increased if the concerned specific factors release its restriction. In other words, we have already figured out the macro-control system for Shanghai e-business. Only the policies aiming at the service quality measures and implementing effect reform can increase the hereof target e-business performance. The quality and the efficiency all depend on the structure optimization policies.

V. Conclusions

Policy design is the essential of the performance evaluation model proposed in this paper. We propose here the performance evaluation model aiming at the detail characteristics of the certain region’s e-business development phase and reaching the policies guide to increase the e-business performance. In this paper, we have already made it clear that e-business performance is restricted by the main variables such as system quality measures, information quality measures, service quality measures implementing effect and therefore all the policies should be guided to sort them out. The key point is that we must match all the variables during the process of e-business. Otherwise, it would go the way unexpected. Actually, the only aim of e-business development should focus on the continuous development by the matching of all specific variables instead of the simplex increase of one or two of its main variables.