Erkie Asmare 1 and Andualem Begashaw

Disadvantage Nonparametric Methods of Efficiency 
Analysis
Besides their benefits, nonparametric methods of efficiency 
analysis also have some draw backs. The first one is this approach 

---

Open Acc Biostat Bioinform
Copyright © 
Erkie Asmare Beyene
4/7
How to cite this article:
Erkie A, Andualem B. Review on Parametric and Nonparametric Methods of Efficiency Analysis. Open Acc Biostat Bioinform
. 2(2). 
OABB.000534. 2018. DOI:
10.31031/OABB.2018.02.000534
Volume 2 - Issue - 2
does not have a solid statistical foundation behind it and is sensi-
tive to outliers [19]. In addition, nonparametric efficiency, repre
-
sented by conventional performance ratios is influenced by input 
and output prices, firm sizes and other exogenous factors, which 
restrain the ratios from reaching closer estimates of the manag-
ers’ true performance [6]. It does not also assume any particular 
functional form for the frontier or the distribution of inefficiency 
[1]. Therefore, does not allow for random error due to, for example, 
measurement error, good or bad luck, miss-specifying inputs and 
outputs, weather, strikes and the like. Similarly, a study by Toma P 
et al. [3] also reported that non-parametric methods of efficiency 
analysis do not take into account the uncertainty characterizing the 
real world (so-called stochastic error).
A study by Kuosmanen et al. [1] and Murillo Zamorano & Vega 
Cervera [4] also asserted that nonparametric methods of efficien
-
cy analysis estimator of model is based on the assumption of no 
noise (i.e., v
i
=0 for all firms i). Thus, any deviation from the fron
-
tier is forced to be attributed to inefficiency (Wang & Wang, 2002).
Because of nonparametric methods does not distinguish between 
inefficiency and statistical noise effects [4], the full distance of a 
brand to the efficiency frontier is interpreted as inefficiency. Sim
-
ilarly, since the nonparametric model is non-stochastic, noise is re-
ported as inefficiency, hence a lower mean technical efficiency [17]. 
But a measurement error or other noise and outliers may influence 
the shape and position of the frontier [10].
According to Ajibefun [2] nonparametric methods couldn’t 
estimate parameters for the model and hence impossible to test 
hypothesis concerning the performance of the model. Similarly, 
a study by Porcelli [16] confirmed that nonparametric efficiency 
analysis does not allow for measurement errors and confidence in
-
tervals [3] hence, it lacks statistical inference. Moreover, nonpara-
metric methods of efficiency are non-stochastic; consequently, effi
-
ciency scores are contaminated by omitted variables, measurement 
error, and other sources of statistical noise [16]. Furthermore, non-
parametric statistical inference often assumes no knowledge about 
the distribution of the underlying population. Therefore, nonpara-
metric inference has deficiency in the sense that it does not utilize 
all the information in the sample and, thus, will be less efficient 
than parametric inference [17].