Sie sind auf Seite 1von 4

AN INPUT DISTANCE FUNCTION APPROACH TO THE MEASUREMENT OF TECHNICAL AND ALLOCATIVE EFFICIENCY: WITH APPLICATION TO INDIAN DAIRY PROCESSING

PLANTS 1. Introduct on
The measurement of the cost efficiency of a firm, and its decomposition into technical and allocative components, has been the subject of a number of papers since the seminal work of Farrell (1957 !Farrell showed how these various efficiency components could be measured, "iven that one had an estimate of the production technolo"y .#any possible estimation methods have been proposed over the years! The two most popular methods have been the stochastic frontier analysis ($F% method proposed by %i"ner, &ovell and $chmidt (1977 and #eeusen and van den 'roeck (1977 , and the data envelopment analysis (()% method, developed by 'oles (19** , %friat (197+ and ,harnes, ,ooper and -hodes (197. !$F% involves the use of econometric methods, while ()% is a linear pro"rammin" techni/ue!0n this paper 0 will use the $F% method because we believe that data noise is an important issue, especially in cases when allocative efficiency is of interest!

2.REVIEW OF LITERATURE.
1rior to 1991, dairy processin" in 0ndia involved a mi2ture of private and cooperative plants! 3owever, the industry was dominated by plants that were run as cooperatives! This was a result of a "overnment policy that prevented private operators from settin" up in areas where cooperative plants were established! 3owever, in 1991 the "overnment introduced new re"ulations that made it easier for private companies to establish processin" plants! This was done to stimulate improved efficiency in the dairy processin" industry! The chan"e in re"ulations resulted in the establishment of over 144 new private plants in the 19915199+ period!15 The purpose of this empirical analysis is to assess whether private plants are better or cooperative!

!. M"t#odo$o%&
The production technolo"y of a dairy plant may be described usin" input sets, &(y , representin" the set of all (671 input vectors, 2, which can be used to produce the (#71 output vector, y! That is, &(y 8 {29 2 can produce y}! The input distance function, d(2,y , is then defined on this input set as9

d'()&* + ,u-{ : '(. * L'&*} )

where is a positive scalar! The distance function, d(2,y , is non:decreasin", positively linearly homo"eneous and concave in 2, and non:increasin" and /uasiconcave in y! The value of the distance will be e/ual to one or "reater than one if the input vector, 2, is an element of the feasible input set, &(y , ie! d(2,y 1 if 2&(y !The distance function has a value of unity if 2 is located on the inner boundary of the input set! The lo" form of a ,obb:(ou"las input distance function, for the case of one output, 6 inputs, ; firms and T time periods, is specified as

where yit is the output /uantity, 2jit is the j:th input /uantity, ln represents a natural lo"arithm, and , and j are unknown parameters to be estimated!0mposin" the restriction for homo"eneity of de"ree <1 in inputs upon this function we obtain the estimatin" e/uation9

=bserve that in the above e/uation we have defined >ln d it 8 it 8 vit:uit to indicate that the distance term may be interpreted as a traditional $F% disturbance term! That is, the distances in a distance function (which are the radial distances between the data points and the frontier could be due to either noise (vit or technical inefficiency (uit ! This is the standard $F% error structure! 0 assume that the vit are i!i!d! ;(4,v+ and uit8ui?e2p((T:t @, where ui is i!i!d! A;(,u+ A! Biven these distributional assumptions, the values for unknown parameters can be obtained by the method of ma2imum likelihood!14 The input:orientated technical efficiency (T) scores are then predicted usin" the conditional e2pectation predictor9 T)it 8 )?(e2p(:uit |it @, (5 proposed by 'attese and ,oelli (199+ ! =nce the parameters of the ,obb:(ou"las input distance function have been estimated, one can derive the correspondin" parameters of the dual cost function! The ,obb:(ou"las cost function is defined as9

where cit is the cost of production, pjit is the j:th input price, and b4, bj and a1 are unknown parameters! Csin" the first order conditions for cost minimisation,

the parameters of the cost and input distance functions are related as follows9

=nce parameters have been of the input distance function have been estimated, one can predict the technical efficiency scores ! =ne can then predict the technically efficient input /uantities as9

The cost:efficient input /uantities are predicted by makin" use of $hephardDs &emma, which states that they will e/ual the first partial derivatives of the cost function9

where cEit is the cost prediction !Thus, for a "iven level of output, the minimum cost of production is it , 2E it !p ,1+ while the observed cost of operation of the firm is 2 it !pit ! These two cost measures are then used to calculate the cost efficiency (,) scores for the i:th firm in the t:th year!

Then, followin" Farrell (1957 , allocative efficiency (%) can be calculated residually as9

)ach of these three efficiency measures take a value between Fero and one, with a value of one indicatin" full efficiency!

3. HYPOTHESIS
The private plants more efficient than the cooperative plants

/.DATA SOURCE
1!1-=G))$ (%T%'%$)! +! 0;(0%$T%T!,=#

0. CONCLUSION
%t this sta"e it is difficult to conclude because calculations have not been done!'ut seein" the proffesionalism in private sector,it can be concluded that privat sector is better than public sector!

7.REFERENCE $in"h, $!$! (+444 , Performance of Dairy Cooperatives and Private Dairies at the Milk Plant Level in India, 1h( Thesis, Cniversity of ;ew )n"land, %rmidale, %ustralia!
$chmidt, 1! and ,!%!6! &ovell (1979 , H)stimatin" Technical and %llocative 0nefficiency -elative to $tochastic 1roduction and ,ost Frontiers I!

Gikipedia!

Das könnte Ihnen auch gefallen