Scale efficiency and homotheticity: equivalence of primal and dual measures

Abstract

We address the issue of equivalence of primal and dual measures of scale efficiency in general production theory framework. We find that particular types of homotheticity of technologies, which we refer to here as scale homotheticity, provide necessary and sufficient condition for such equivalence. We also identify the case when the scale homotheticity is equivalent to the homothetic structures from Shephard (Theory of cost and production functions, Princeton studies in mathematical economics. Princeton University Press, Princeton, 1970).

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

Fig. 1
Fig. 2
Fig. 3

Notes

  1. 1.

    E.g., see Färe et al. (1986), Färe and Primont (1995), Zelenyuk (2011, 2013a, b) for details.

  2. 2.

    For the properties of the Shephard’s distance function see Shephard (1953, 1970).

  3. 3.

    The origins of this measure go back to at least Førsund and Hjalmarsson (1979). For this and other ways of measuring scale issues see, for example, Banker et al. (1984), Färe and Grosskopf (1985), Färe et al. (1986), Førsund (1996), Zelenyuk (2011, 2013a, b) to mention just a few.

  4. 4.

    To be more precise, q and f must also satisfy the following properties: q(0) = 0; q(y) > 0 if \(y\geq0; q(y^{n})\rightarrow+\infty\) for \(\left\{\|y^{n}\| \right\} \rightarrow+\infty; f(0)=0; \) and for any x > 0 or x ≥ 0 such that fx) > 0, for some scalar δ > 0, we have \(f(\delta x)\rightarrow+\infty\) as \(\delta\rightarrow+\infty.\) To preserve convexity of L(y), we also need q(y) being quasi-convex on \({{\mathbb{R}}_{+}^{M}}\) and f(x) being quasi-concave on \({{\mathbb{R}}_{+}^{N}.}\) Also, the transform \({F:{\mathbb{R}}_{+}\rightarrow{\mathbb{R}}_{+}}\) means it is a finite real-valued non-negative, upper semi-continuous and non-decreasing function such that F(0) = 0. See Shephard (1970) for details of these properties.

  5. 5.

    To be precise, \(\Uppsi(z)=\min\{v: F(v)\geq z\}, z\geq 0,\) and so it is a finite real-valued non-negative, lower semi-continuous and non-decreasing function such that \(\Uppsi(0)=0\) and \(\Uppsi(z)>0,\) for z > 0.

  6. 6.

    More precisely, note that \(c(w)=\min_{x}\left\{ wx: f(x)\geq 1\right\}.\)

  7. 7.

    I thank anonymous referee for hinting me towards establishing this more general result than the one I had in the first version of this paper.

  8. 8.

    E.g., see recent works of Simar and Wilson (2007, 2011) and Simar and Zelenyuk (2006, 2007, 2011) and references cited therein.

  9. 9.

    See Shephard (1970) for the proofs of these and other properties of D i (yx) and of C(yw) that we mention here.

References

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

    Article  Google Scholar 

  2. Chambers RG, Färe R (1994) Hicks neutrality and trade biased growth: a taxonomy. J Econ Theory 64(2):554–567

    Article  Google Scholar 

  3. Chambers RG, Färe R (1998) Translation homotheticity. Econ Theory 11(3):629–641

    Article  Google Scholar 

  4. Färe R, Grosskopf S (1985) A nonparametric cost approach to scale efficiency. Scand J Econ 87(4):594–604

    Article  Google Scholar 

  5. Färe R, Grosskopf S (2000) Notes on some inequalities in economics. Econ Theory 15(1):227–233

    Article  Google Scholar 

  6. Färe R, Li SK (2001) A nonparametric test of translation homotheticity. Econ Lett 72(3):341–345

    Article  Google Scholar 

  7. Färe R, Primont D (1995) Multi-output production and duality: theory and applications. Kluwer Academic Publishers, Boston

    Google Scholar 

  8. Färe R, Grosskopf S, Lovell CAK (1986) Scale economies and duality. J Econ 46:175–182 doi:10.1007/BF01229228

    Article  Google Scholar 

  9. Färe R, Grosskopf S, Lovell C (1994) Production frontiers. Cambridge University Press, Cambridge

    Google Scholar 

  10. Färe R, Førsund FR, Grosskopf S, Hayes K, Heshmati A (2001) A note on decomposing the Malmquist productivity index by means of subvector homotheticity. Econ Theory 17(1):239–245

    Article  Google Scholar 

  11. Førsund FR (1996) On the calculation of the scale elasticity in DEA models. J Prod Anal 7(2–3):283–302. doi:10.1007/BF00157045.

    Article  Google Scholar 

  12. Førsund FR, Hjalmarsson L (1979) Generalised farrell measures of efficiency: an application to milk processing in swedish dairy plants. Econ J 89(354):294–315

    Article  Google Scholar 

  13. Peyrache A (2013) Multilateral productivity comparisons and homotheticity. J Prod Anal 40(1):57–65

    Google Scholar 

  14. Seitz WD (1970) The measurement of efficiency relative to a frontier production function. Am J Agric Econ 52(4):505–511

    Article  Google Scholar 

  15. Shephard R (1970) Theory of cost and production functions, Princeton studies in mathematical economics. Princeton University Press, Princeton

    Google Scholar 

  16. Shephard RW (1953) Cost and production functions. Princeton University Press, Princeton

    Google Scholar 

  17. Simar L, Wilson P (2011) Inference by the m out of n bootstrap in nonparametric frontier models. J Prod Anal 36(1):33–53. doi:10.1007/s11123-010-0200-4

    Article  Google Scholar 

  18. Simar L, Wilson PW (2007) The measurement of productive efficiency, chap. Statistical inference in nonparametric frontier models: recent developments and perspectives, 2 edn. Oxford University Press, pp. 421–521

  19. Simar L, Zelenyuk V (2006) On testing equality of distributions of technical efficiency scores. Econom Rev 25(4):497–522

    Article  Google Scholar 

  20. Simar L, Zelenyuk V (2007) Statistical inference for aggregates of farrell-type efficiencies. J Appl Econ 22(7):1367–1394

    Article  Google Scholar 

  21. Simar L, Zelenyuk V (2011) Stochastic fdh/dea estimators for frontier analysis. J Prod Anal 36(1):1–20

    Article  Google Scholar 

  22. Zelenyuk V (2011) Scale efficiency: equivalence of primal and dual measures. CEPA working papers series WP09/2011, University of Queensland, School of Economics

  23. Zelenyuk V (2013a) A note on equivalences in measuring returns to scale. Int J Bus Econ 12(1) (in press)

  24. Zelenyuk V (2013b) A scale elasticity measure for directional distance function and its dual: theory and DEA estimation. Eur J Oper Res 228(3):592–600

    Google Scholar 

Download references

Acknowledgments

I would like to thank Erwin Diewert, Rolf Färe, John Farrell, Shawna Grosskopf, Knox Lovell, Chris O’Donnell, Antonio Peyrache and Prasada Rao for their valuable comments. I also thank the audience where this paper was presented and, in particular, participants of my seminar at the University of Auckland and at the Productivity Workshop of CEPA. I especially thank anonymous referees and the editor for their valuable comments; they helped improving this paper substantially. I remain to be solely responsible for my views expressed in this article.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Valentin Zelenyuk.

Appendix

Appendix

Proof of Proposition 1

Assume a regular technology given by \({L(y), y\in{\mathbb{R}}_{+}^{M}}\) is input scale homothetic, i.e., \({L(y)=G(y)L(y|crs), y\in{\mathbb{R}}_{+}^{M}, G(y)\geq1,}\) then \({\forall(x,y)\in{\mathbb{R}}_{+}^{N+M}}\) we have

$$ \begin{aligned} D_{i}(y,x) & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}: x/\lambda\in L(y)\right\}\\ & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}: x/\lambda\in G(y)L(y|crs)\right\}\\ & = (G(y))^{-1}\sup_{\lambda}\left\{ \lambda G(y)\in{{\mathbb{R}}}_{++}:\left(\frac{x}{\lambda G(y)}\right)\in L(y|crs)\right\} \\ & = (G(y))^{-1}\sup_{\hat{\lambda}}\left\{ \hat{\lambda}\in{{\mathbb{R}}}_{++}: x/\hat{\lambda}\in L(y|crs)\right\} (\hbox{where }\hat{\lambda}=\lambda G(y), y\in{{\mathbb{R}}}_{+}^{M}) \\ & = (G(y))^{-1}D_{i}(y,x|crs). \\ \end{aligned} $$

To prove the converse, assume that for a regular technology characterized by L(y) we have \({D_{i}(y,x)=(G(y))^{-1}D_{i}(y,x|crs), \forall(x,y)\in{\mathbb{R}}_{+}^{N+M},}\) for some finite real-valued, lower semi-continuous function \( {G:{\mathbb{R}}_{+}^{M}\rightarrow [1,\infty)} \). Then it must be also true that for all \({y\in{\mathbb{R}}_{+}^{M}}\) we also have

$$ \begin{aligned} L(y) &= \left\{ x\in{{\mathbb{R}}}_{+}^{N}: D_{i}(y,x)\geq1\right\} \\ &= \left\{ x\in{{\mathbb{R}}}_{+}^{N}:(G(y))^{-1}D_{i}(y,x|crs)\geq1\right\} \\ &= G(y)\left\{ x/G(y)\in{{\mathbb{R}}}_{+}^{N}: D_{i}(y,x/G(y)|crs)\geq1\right\} \\ &= G(y)\left\{ \hat{x}\in{{\mathbb{R}}}_{+}^{N}: D_{i}(y,\hat{x}|crs)\geq1\right\} \\ &= G(y)L(y|crs),\\ \end{aligned} $$

where we used the complete characterization property (2.4), the linear homogeneity property of D i (yx), namely that D i (y, α x|crs) = D i (yx|crs)α, ∀α > 0 and made \({\hat{x}=x/G(y), y\in{\mathbb{R}}_{+}^{M}.}\) \(\square\)

Proof of Proposition 2

Assume a regular technology characterized by \({L(y), y\in{\mathbb{R}}_{+}^{M}}\) is also ISH, i.e., L(y) = G(y)L(y|crs), for some G(y) ≥ 1 and \({\forall y\in{\mathbb{R}}_{+}^{M},}\) then for all \({w\in{\mathbb{R}}_{++}^{N}}\)and for all \({y\in{\mathbb{R}}_{+}^{M}}\) such that L(y) ≠ ∅ we must also have

$$ \begin{aligned} C(y,w) & = \mathop{\hbox{min}}\limits_{x}\{wx: x\in L(y)\} \\ & = \mathop{\hbox{min}}\limits_{x}\left\{ wx: x\in G(y)L(y|crs)\right\} \\ & = G(y)\mathop{\hbox{min}}\limits_{x}\left\{ w\frac{x}{G(y)}:\frac{x}{G(y)}\in L(y|crs)\right\} \\ & = G(y)\min_{\hat{x}}\left\{ w\hat{x}:\hat{x}\in L(y|crs)\right\} (\hbox{where } \hat{x}=x/G(y),y\in{{\mathbb{R}}}_{+}^{M})\\ & = G(y)C(y,w|crs).\\ \end{aligned} $$

To prove the converse, assume that for a regular technology characterized by \({L(y), y\in{\mathbb{R}}_{+}^{M}}\) we also have C(yw) = G(y)C(yw|crs) for some \({G(y)\geq1, \forall(x,y)\in{\mathbb{R}}_{+}^{N+M},}\) and assume that L(y) is convex, then from the duality theory result (2.11), it must be true that for all \({y\in{\mathbb{R}}_{+}^{M}}\) we also have

$$ \begin{aligned} L(y) & = \left\{ x\in{{\mathbb{R}}}_{+}^{N}: wx \ge C(y,w),\forall w\in{{\mathbb{R}}}_{++}^{N}\right\} \\ & = \left\{ x\in{{\mathbb{R}}}_{+}^{N}: wx \ge G(y)C(y,w|crs),\forall w\in{{\mathbb{R}}}_{++}^{N}\right\} \\ & = G(y)\left\{ \frac{x}{G(y)}\in{{\mathbb{R}}}_{+}^{N}: w\frac{x}{G(y)} \ge C(y,w|crs),\forall w\in{{\mathbb{R}}}_{++}^{N}\right\} \\ & = G(y)\left\{ \hat{x}\in{{\mathbb{R}}}_{+}^{N}: w\hat{x} \ge C(y,w|crs),\forall w\in{{\mathbb{R}}}_{++}^{N}\right\} (\hbox{where } \hat{x}=x/G(y),y\in{{\mathbb{R}}}_{+}^{M}) \\ & = G(y)L(y|crs). \\ \end{aligned} $$

Note that convexity of L(y) is only required to prove the converse part (i.e., the necessity of ISH), and C(yw) = G(y)C(yw|crs) for some G(y) ≥ 1 and \({\forall(x,y)\in{\mathbb{R}}_{+}^{N+M}}\) is ensured by ISH whether L(y) is convex or not, for any regular technology. \(\square\)

Proof of Proposition 3

Necessity: Suppose technology satisfies input scale homotheticity, then (and only then), due to (3.4), we have \({S_{i}(y,x)=S_{c}(y,w), \forall(x,y)\in{\mathbb{R}}_{+}^{N+M},}\) which in turn implies that S i (yx) must be independent of x, while S c (yw) must be independent of w. This means that there exist some functions ϕ(x), ψ(x), h(y), and \(\check{h}(y)\) such that D i (yx) = ϕ(x)/h(y) and \(D_{i}(y,x|crs)=\phi(x)/\check{h}(y)\) as well as C(yw) = ψ(w)h(y) and \(C(y,w|crs)=\psi(w)\check{h}(y)\) and so that \(SE_{i}(y,x)=\check{h}(y)/h(y)=SE_{c}(y,w).\)

What is left is to show that ϕ(x), h(y), ψ(w) and \(\check{h}(y)\) satisfy properties required by the definition of homothetic input structure. Let us first show that ϕ(x) is a upper semi-continuous, linearly homogeneous and non-decreasing function such that ϕ(0) = 0, and for any x ≥ 0 such that ϕ(δx) > 0, for some scalar δ > 0, we have \(\phi(\delta x)\rightarrow+\infty\) as \(\delta \rightarrow+\infty.\) To do so, note that D i (yx) is a finite real-valued continuous and linearly homogeneous in x on \({{\mathbb{R}}_{+}^{N}}\) and therefore because D i (yx) = ϕ(x)/h(y) these same properties must also be shared by ϕ(x). Footnote 9 Similarly, because D i (y, 0) = 0, we also have ϕ(0) = 0. Moreover, for any x ≥ 0 such that ϕ(δx) > 0, for some scalar δ > 0, we will have \(\phi(\delta x)\rightarrow+\infty\) as \(\delta\rightarrow+\infty,\) because for any x ≥ 0 such that D i (y,) > 0 we have \(D_{i}(y,x\delta)\rightarrow0\) when \(\delta\rightarrow+\infty.\)

Furthermore, let us show that h(y) is a finite real-valued lower semi-continuous non-decreasing function such that h(0) = 0, and h(y) > 0 if y ≥ 0, while \(h(y^{n})\rightarrow+\infty\) for \(\left\{\|y^{n}\|\right\} \rightarrow+\infty.\) To do so, recall that C(yw) is a finite, lower semi-continuous, and non-decreasing in y on \({{\mathbb{R}}_{+}^{M},}\) and so, because C(yw) = ψ(w)h(y), the same properties should hold for h(y). Also, because D i (y,x) > 0 when y ≥ 0, it must be true that h(y) > 0 if y ≥ 0, since we have D i (yx) = ϕ(x)/h(y). Similarly, because \(D_{i}(y,x)\rightarrow+\infty\) for \(y\rightarrow0,\) we have \(h(y)\rightarrow0,\) when \(y\rightarrow0.\) Analogously, \(h(y^{n})\rightarrow+\infty\) for \(\left\{\|y^{n}\|\right\} \rightarrow+\infty.\)

Next, since we involve convexity of L(y), we need to show that h(y) is quasi-convex on \({{\mathbb{R}}_{+}^{M}}\) and ϕ(x) is quasi-concave on \({{\mathbb{R}}_{+}^{N}.}\) These are established by noting that D i (yx) is concave and non-decreasing in x on \({{\mathbb{R}}_{+}^{N}}\) and is quasi-concave in y on \({{\mathbb{R}}_{+}^{M},}\) which in turn implies that, to ensure that D i (yx) = ϕ(x)/h(y) holds, ϕ(x) must be quasi-concave on \({{\mathbb{R}}_{+}^{N},}\) while 1/h(y) is quasi-concave on \({{\mathbb{R}}_{+}^{M}}\) and so h(y) is quasi-convex on \({{\mathbb{R}}_{+}^{M}.}\) (One may also establish similar properties of ψ(x) and \(\check{h}(y)\) by a similar argument).

Combining all these conclusions, we conclude that (4.2) is satisfied, with all the required properties for the involved functions, and so technology (and its CRS-hypothetical counterpart) has homothetic input structure.

Sufficiency: If technology is regular and satisfies \(D_{i}(y,x|crs)<+\infty\) for any \((x,y)\in T\) where y ≥ 0, then \(C(y,w|crs)<+\infty\) for any \((x,y)\in T\) where \(D_{i}(y,x)<+\infty\) and \(C(y,w)<+\infty.\) This ensures existence of SE i (yx) and SE c (yw). Now, suppose technology has input homothetic structure defined in (4.1). Then, (4.2) and (4.3) must hold and, moreover, we must also have

$$ D_{i}(y,x|crs) = f(x)/\check{q}(y)\\ $$
(6.1)

and

$$ C(y,w|crs) = c(w)\check{q}(y). $$
(6.2)

where \({\check{q}(y)=\inf_{\delta}\left\{ \delta\Uppsi(q(y/\delta)),\forall\delta \in{\mathbb{R}}_{++}\right\}, \Uppsi(z)=\min\{v: F(v)\geq z\}, z\geq0}\) and \(c(w)=\min_{x}\left\{ wx: f(x)\geq 1\right\}.\) Specifically, (6.1) is true because

$$ \begin{aligned} D_{i}(y,x|crs) & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}:(x/\lambda,y)\in\check{T}\right\} \\ & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}:(x/\lambda,y)\in\delta T,\forall\delta\in{{\mathbb{R}}}_{++}\right\} \\ & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}: F\left(f\left(x/(\lambda\delta)\right)\right)\geq q(y/\delta),\forall\delta\in{{\mathbb{R}}}_{++}\right\} \\ & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}: f\left(x/(\lambda\delta)\right)\geq\Uppsi(q(y/\delta)),\forall\delta\in{{\mathbb{R}}}_{++}\right\} \\ & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}: f\left(x\right)/(\lambda\delta)\geq\Uppsi(q(y/\delta)),\forall\delta\in{{\mathbb{R}}}_{++}\right\} \\ & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}: f(x)/\left(\delta\Uppsi(q(y/\delta))\right)\geq\lambda,\forall\delta\in{{\mathbb{R}}}_{++}\right\} \\ & = \sup_{\lambda}\left\{ \lambda\in{{\mathbb{R}}}_{++}: f(x)/\inf_{\delta}\left\{ \delta\Uppsi(q(y/\delta)),\forall\delta\in{{\mathbb{R}}}_{++}\right\} \geq\lambda\right\} \\ & = \frac{f(x)}{\check{q}(y)},\left(\hbox{ where } \check{q}(y)=\inf_{\delta}\left\{ \delta\Uppsi(q(y/\delta)),\forall\delta\in{{\mathbb{R}}}_{++}\right\} \right). \\ \end{aligned} $$

and, in turn, (6.2) is true because

$$ \begin{aligned} C(y,w|crs) & = \mathop{\hbox{min}}\limits_{x}\{wx:(x,y)\in\check{T}\} \\ & = \mathop{\hbox{min}}\limits_{x}\{wx: D_{i}(y,x|crs)\geq 1\} \\ & = \mathop{\hbox{min}}\limits_{x}\{wx:\frac{f(x)}{\check{q}(y)}\geq 1\} \\ & = \check{q}(y)\mathop{\hbox{min}}\limits_{x}\{wx/\check{q}(y): f(x/\check{q}(y))\geq 1\}\\ & = \check{q}(y)\min_{\hat{x}}\left\{ w\hat{x}: f(\hat{x})\geq 1\right\} (\hbox{where } \hat{x}=x/\check{q}(y), y\in{{\mathbb{R}}}_{+}^{M}) \\ & = \check{q}(y)c(w),\left(\hbox{where}\,c(w)=\min_{\hat{x}}\left\{ w\hat{x}: f(\hat{x})\geq1\right\} \right). \\ \end{aligned} $$

Now, combining (6.1) with (4.2) and (6.2) with (4.3), we get

$$ SE_{i}(y,x)=\frac{D_{i}(y,x)}{D_{i}(y,x|crs)}= \frac{\check{q}(y)}{q(y)}=\frac{C_{i}(y,w|crs)}{C_{i}(y,w)}= SE_{c}(y,w) $$

which in turn implies technology is input scale homothetic, with \(G(y)=\Uppsi(q(y))/\check{q}(y)\) satisfying the required properties. Indeed, because \({\check{q}(y)=\inf_{\delta}\left\{ \delta\Uppsi(q(y/\delta)),\forall \delta\in{\mathbb{R}}_{++}\right\} }\) we always have \({\check{q}(y)\leq\Uppsi(q(y)),\;\forall y\in{\mathbb{R}}_{+}^{M},}\) thus ensuring that G(y) ≥ 1. Finally, note that q(y) is a finite real-valued lower semi-continuous on \({{\mathbb{R}}_{+}^{M},}\) while \(\Uppsi(\cdot)\) is a finite real-valued lower semi-continuous on \({{\mathbb{R}}_{+}}\) and so these same properties must be also shared by \(\check{q}(y),\) which was defined as \({\check{q}(y)= \inf_{\delta}\left\{ \delta\Uppsi(q(y/\delta)), \forall\delta\in{\mathbb{R}}_{++}\right\}.}\) All these, in turn, ensure that \(G(y)=\Uppsi(q(y))/\check{q}(y)\) is also a finite real-valued lower semi-continuous function on \([1,+\infty).\) \(\square\)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Zelenyuk, V. Scale efficiency and homotheticity: equivalence of primal and dual measures. J Prod Anal 42, 15–24 (2014). https://doi.org/10.1007/s11123-013-0361-z

Download citation

Keywords

  • Production theory
  • Scale efficiency
  • Homotheticity
  • Duality theory

JEL Classification

  • D24