Professional Documents
Culture Documents
FGNZ Reply
FGNZ Reply
FGNZ Reply
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review.
http://www.jstor.org
average product (consistent with the minimum Perhaps the most controversial component
point on a long-run, U-shaped average cost (due perhaps to our exposition) in the FGNZ
curve) which provides a very nice benchmark decomposition is the scale change component.
for identifying both optimal scale in the one The discussion of identification and compu-
period case (also pointed out by RD), and tation of scale efficiency in the single-period
technical change in the multiperiod case (as case in RD is useful here. They provide an
we discuss below). intuitive discussion of the role of the CRS
Turning to the decomposition in RD versus technology as a benchmarkin identifying what
FGNZ, we note that the VRS efficiency Rajiv D. Banker (1984) calls the most pro-
change term is identical in the two decompo- ductive scale size (MPSS), and how that may
sitions, so we turn to the remaining two com- be used to identify deviations from that opti-
ponents: technical change and scale change. mal size in a given period; for example, from
The technical change term used in both RD equation (8) in RD, scale efficiency would be
and FGNZ can be summarized as computed in period "0" as
interpretationof the scale change component tical in RD and FGNZ; in that sense, both are
is whether an observation got closer to or far- "internally consistent." Both RD and FGNZ
ther from most productive scale size between rely on both CRS and VRS technologies to
the two periods, where MPSS may differ from identify their components of productivity
period to period. In contrast, RD define scale change, although the specification and inter-
change as pretation of the technical change and scale
change components are different, as are, of
(5) SCHRD course, the empirical values of these compo-
nents. Computationally, the RD scale and
= SE1(xl, yl) SE0(xl, yl) 11/2 technical change terms may not always be
computable; the corresponding FGNZ terms
[SEo(xo, yo) SEl(x0,y0) J
are computable.
One of the messages of RD is that there is
The first ratio in the squarebracketsis iden- more than one way to decompose an index.
tical to the FGNZ scale component, and the We would like to point out two other varia-
second ratio includes what look like scale ef- tions for the interested reader.Based on an al-
ficiency terms defined across periods. The sec- ternative distance function definition of
ond ratio thus requires the computation of productivity which W. Erwin Diewert ( 1992)
distance functions like D?(xj, Yl), in which dubbed the "Hicks-Moorsteen" index, Hans
data from one period ( " 1") are evaluated rel- Bjurek (1994) derives a decomposition into
ative to a VRS benchmark from the previous technical change and efficiency change. Like
period. Again, as seen in the case of Ireland, the FGNZ decomposition, the components of
this can cause computationalproblems. These this index are not susceptible to nonexistence
problems do not arise in the FGNZ decom- of solutions to the individual distance func-
position precisely because it contains no tion problems.8C. A. Knox Lovell and Emili
mixed-period terms computed under VRS. Grifell-Tatje ( 1994) propose a similar form of
Finally, the RD scale term may incorrectly productivity index using distance functions,
identify the scale properties of the underlying with a slightly different decomposition.
technology.7 Perhaps the main contribution of RD is to
To sum up, RD provide an alternative spec- raise the issue of the interpretationof the role
ification of the FGNZ decomposition of the of the reference technology in computing pro-
Malmquist productivity measure. The result- ductivity. Perhapsmost importantly,they have
ing overall measure of productivity is iden- confirmed that the correct benchmarkin com-
puting overall Malmquist productivity (if it is
to be a measure of TFP) is a CRS benchmark,
a source of much confusion in the literature.
7 As the following example shows, the RD scale com- In pointing out how to interpretthe use of CRS
ponent may reportvalues that do not reflect the scale prop-
erties of the data. Suppose we have two firms, A and B, and VRS technologies in the computation of
which use one input to produce a single output. scale efficiency, for example, they are, in ef-
fect, saying that we are using these technolo-
Firm (xo, yo) (xI, y1) gies as benchmarks rather than arguing, for
A (6,3) (5,5) example, that the true technology is CRS. In
fact we agree with this general viewpoint-
B (4,2) (3, 3) the "true" technology is unknown; however,
thanks to the techniques we use, we can con-
Based on this sample, both firms are producing at a point
of CRS in each period, t = 0, 1. Since both firms were at struct useful benchmarks based on observed
MPSS in both periods, the FGNZ scale component reg- data to use in identifying productivity growth
isters a value of one, reflecting no change in scale effi- and its sources.
ciency, as it should. The RD decomposition gives a scale
change value for firm A of (2/1.66667)/2, which is ap-
proximately 1.095, suggesting an improvement in scale
efficiency, which did not occur. For firm B in the RD
decomposition, there is no solution. 8 For a more detailed discussion, see Fare et al. ( 1996).