Skip to main content

The extension and integration of the inverse DEA method


The inverse DEA (Data Envelopment Analysis) method is primarily used to analyse the changing relationship between the inputs and outputs of a DMU (Decision-Making Unit) when its efficiency is kept constant or set to a target value. However, the existing inverse DEA method cannot be applied directly to estimate all the changing relationships. For example, the existing DEA models fail to estimate the input variations when the supervisor wants to maintain the DMU’s output-oriented efficiency during the downscaling of production. This paper analyses all the possible changing relationships that need to be solved by the inverse DEA method and develops different models for both the output and input orientations, accomplishing the extension and integration of the inverse DEA model. For illustration of our results, a numerical example is given.

This is a preview of subscription content, access via your institution.

Figure 1
Figure 2
Figure 3


  1. 1.

    This lemma has been discussed in Dulá and Thrall (2001). In this paper, we present and prove this lemma in a completely new way, which is more close to our topic and is the foundation for proving Theorems 1 and 2.


  1. Amin GR and Emrouznejad A (2007). Inverse linear programming in DEA. International Journal of Operations Research 4(2): 105–109.

    Google Scholar 

  2. Banker RD, Charnes A and Cooper WW (1984). Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Science 30(9): 1078–1092.

    Article  Google Scholar 

  3. Charnes A, Cooper WW and Rhodes E (1978). Measuring the efficiency of decision making units. European Journal of Operational Research 2(6): 429–444.

    Article  Google Scholar 

  4. Dulá JH and Thrall RM (2001). A computational framework for accelerating DEA. Journal of Productivity Analysis 16(1): 63–78.

    Article  Google Scholar 

  5. Färe R and Grosskopf S (1985). A nonparametric cost approach to scale efficiency. The Scandinavian Journal of Economics 87(4): 594–604.

    Article  Google Scholar 

  6. Gattoufi S, Amin GR and Emrouznejad A (2014). A new inverse DEA method for merging banks. IMA Journal of Management Mathematics 25(1): 73–87.

    Article  Google Scholar 

  7. Hadi-Vencheh A and Foroughi AA (2006). A generalized DEA model for inputs/outputs estimation. Mathematical and Computer Modelling 43(5): 447–457.

    Article  Google Scholar 

  8. Han S (2003). An inverse DEA model with stochastic factors. Mathematics in Practice and Theory 33(03): 23–29.

    Google Scholar 

  9. Jahanshahloo GR, Vencheh AH, Foroughi AA and Kazemi Matin R (2004a). Inputs/outputs estimation in DEA when some factors are undesirable. Applied Mathematics and Computation 156(1): 19–32.

    Article  Google Scholar 

  10. Jahanshahloo GR, Lotfi FH, Shoja N, Tohidi G and Razavyan S (2004b). Input estimation and identification of extra inputs in inverse DEA models. Applied Mathematics and Computation 156(2): 427–437.

    Article  Google Scholar 

  11. Jahanshahloo GR, Lotfi FH, Shoja N, Tohidi G and Razavyan S (2004c). The outputs estimation of a DMU according to improvement of its efficiency. Applied Mathematics and Computation 147(2): 409–413.

    Article  Google Scholar 

  12. Lertworasirikul S, Charnsethikul P and Fang S-C (2011). Inverse data envelopment analysis model to preserve relative efficiency values: The case of variable returns to scale. Computers & Industrial Engineering 61(4): 1017–1023.

    Article  Google Scholar 

  13. Lertworasirikul S, Fang S-C, Nuttle HLW and Joines JA (2003). Fuzzy BCC model for data envelopment analysis. Fuzzy Optimization and Decision Making 2(4): 337–358.

    Article  Google Scholar 

  14. Rad AM, Dehghan R and Lotfi FH (2011). Inverse DEA model with fuzzy data for output estimation. Iranian Journal of Optimization 2(2): 388–411.

    Google Scholar 

  15. Seiford LM and Thrall RM (1990). Recent developments in DEA: The mathematical programming approach to frontier analysis. Journal of Econometrics 46(12): 7–38.

    Article  Google Scholar 

  16. Wei QL, Zhang JZ and Zhang XS (2000). An inverse DEA model for inputs/outputs estimate. European Journal of Operational Research 121(1): 151–163.

    Article  Google Scholar 

  17. Yan H, Wei Q and Hao G (2002). DEA models for resource reallocation and production input/output estimation. European Journal of Operational Research 136(1): 19–31.

    Article  Google Scholar 

  18. Zhang XS and Cui JC (1999). A project evaluation system in the state economic information system of China: An operations research practice in public sectors. International Transactions in Operational Research 6(5): 441–452.

    Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Jin-chuan Cui.

Additional information

Supported by Key Laboratory of Management, Decision and Information Systems, Chinese Academy of Science.



Technical proofs

Appendix A.1. Proof of Lemma 1


  • According to the conventional DEA theory, the calculation of the efficiency score of DMU′ o follows the following model:

    Suppose that the optimal solution of model (DLP 0) is (φ*, Λ*, v*) and that of (DLP0) is , then (φ*, Λ*, λ n+1=0, v*) is a feasible solution of (DLP0), which indicates that . Considering constraint (2), we know that φ=1 is a feasible solution of (DLP0). Thus, we have

    Considering model (DLP0), we have λ n+1⩽1 according to constraint (11). Otherwise, we would have ΛX T<0, which contradicts the constraint Λ⩾0. Next, We make a discussion in terms of .

    If , then we have ΛX T=0 and Λ=0. Considering constraint (12), holds. Therefore, .

    If , then make the following transformation:

    We can conclude that is a feasible solution of model (DLP 0), so holds.

    If , then formula (13) indicates that .

    If , then we have

    and , which contradicts formula (13). Therefore, is impossible.

    If , then is a feasible solution of model (DLP 0). Thus, we find that . Therefore, holds due to formula (13).

    To summarise, model (DLP0) has the same optimal solution as model (DLP0). Therefore, when DMU′ o satisfies constraint (2), models (DLP 0) and (DLP0) are equivalent, and both can be used to obtain the efficiency score of DMU′ o. □

Appendix A.2. Proof of Corollary 1


  • We only need to prove that condition (1) always holds when δ 2=1 and δ 3=0.

    Assuming that the Pareto solution S*=(ΔX*o, Λ*, v*) does not meet condition (1), then there exists Λ* that satisfies

    We can conclude that there exists a value τ(0<τ<1), from which we can obtain and satisfying and . The following inequality holds regarding :

    Thus, there exists , and is a feasible solution of model (Output Y), which contradicts the premise that ΔX* o is the Pareto value. □

Appendix A.3. Proof of Corollary 2


  • For the VRS and NDRS efficiency scores of model (Output Y), considering the arbitrary feasible solution , because , we have and

    where Y r=(y r1, y r2, …, y rn)T is an n-dimensional column vector. Thus, the Pareto solution does not meet condition (1). □

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Zhang, M., Cui, Jc. The extension and integration of the inverse DEA method. J Oper Res Soc 67, 1212–1220 (2016).

Download citation


  • data envelopment analysis
  • inverse DEA
  • efficiency
  • production possibility set
  • multi-objective