Sankhya: The Indian Journal of Statistics

1999, Volume 61, Series B, Pt. 2, pp. 337-350

A MODEL OF FPF WITH CORRELATED ERROR COMPONENETS: AN APPLICATION TO INDIAN AGRICULTURE

By

MANORANJAN PAL, *Visva-Bharati, Santiniketan*

and

ATANU SENGUPTA, *Indian Statistical Institute, Calcutta*

*SUMMARY. *The **S**TOCHASTIC **F**RONTIER **P**RODUCTION **F**UNCTION (**SFPF**)
models usually assume that the two
components
of the equation error, namely the noise variable and the inefficiency error
are independently distributed.
In this paper we relax this assumption. This may serve as a basis for
specification test.
Moreover, the marginal distribution of the
inefficiency error is taken as truncated normal instead of half normal. We
develop a model of **SFPF**
introducing both technical and allocative inefficiencies under the
assumption that technical efficiency error and the noise variable are bivariate
normal, the technical inefficiency being truncated at zero. The allocative
inefficiency errors are also correlated among themselves. We obtain the likelihood
function and discuss estimation of technical and allocative inefficiencies.
This is illustrated through an Indian data on agricultural farms.

*AMS (1991) subject classification.* 90A11, 90A19.

*Key words and phrases. * Stochastic frontier production function, technical and
allocative efficiencies, correlated error components.