Sie sind auf Seite 1von 2

Asia Pacific Journal of Management, 19, 587–600, 2002


c 2002 Kluwer Academic Publishers. Manufactured in The Netherlands.

A DEA Approach to Understanding the Productivity


Growth of Malaysia’s Manufacturing Industries
RENUKA MAHADEVAN r.mahadevan@economics.uq.edu.au
Department of Economics, The University of Queensland, Brisbane, Australia

Abstract. This paper seeks to explain the productivity growth performance of Malaysia’s manufacturing sector
using a panel data of 28 industries from 1981–1996. Here, the data envelopment analysis technique is used to
calculate and decompose the Malmquist index of total factor productivity (TFP) growth into technical change,
change in technical efficiency and change in scale efficiency. This allows the identification of the sources of
productivity growth which is crucial for policy formulation. It was found that the annual TFP growth of the
Malaysian manufacturing sector was low at 0.8% and this was driven by small gains in both technical change and
technical efficiency, with industries operating close to optimum scale.

Keywords: total factor productivity growth, technical change, technical efficiency, scale efficiency, data
envelopment analysis

1. Introduction

Before the 1997 financial crisis, Malaysia was often touted to be the next first-tier newly
industrialising economy. The Malaysian economy has not only sustained a GDP growth
rate of 7.7% in the decade of the 70s and 8.8% in 1987–1996, but was also successful in
moving from a natural resource-rich primary industrial-based country in the 1960s to that
of a manufacturing dominant economy by the late 70s. Of late, there has been a clear move
from labour intensive manufacturing operations to mid-level capital intensive manufacturing
operations and this has been significantly helped by the large foreign direct investment (FDI)
that Malaysia has attracted since the mid 80s.
Although Malaysia’s manufacturing output has grown from contributing about 19.3% in
1979 to 34.2% of its GDP output in 1996, the key issue is, how sustainable is this growth?
In this context, one needs to study the total factor productivity (TFP) growth performance
of the manufacturing industries as TFP growth is one of the measures of sustainable output
growth, given the diminishing returns nature of input growth in the long run. The interest
in TFP growth studies on Malaysia’s manufacturing sector has been increasing and some
of the key results are as follows. While the World Bank (1989) showed a TFP growth rate
of −1.9% for 1981–1984, studies such as Okamoto (1994) and Tham (1996, 1997) show
that TFP growth did not exceed 0.3% for the late 80s. Also, a declining trend in the 80s to
mid 90s was evidenced by The Productivity Report 1999, and Mahadevan (2001b).
To date, all but Mahadevan (2001b) has used the stochastic frontier approach to economet-
rically estimate the production function of the Malaysian manufacturing sector to obtain
TFP growth estimates for 28 manufacturing industries. All other studies have used the
588 MAHADEVAN

nonparametric translog index approach for their analysis.1 Here, the data envelopment
analysis (DEA) is used as an alternative technique underlying the nonparametric frontier
approach. The frontier is defined as a set of best obtainable positions obtained as a lo-
cus of constrained maximum or minimum values. Thus an industry which operates on the
production frontier is said to produce its potential or maximum output by following the
‘best practice’ techniques given the technology. In the context of this study, following Färe
et al. (1994), the DEA allows the construction of the ‘best’ frontier based on the data from
the Malaysian manufacturing industries in the sample. Each industry is compared to the
‘model’ or ‘best’ frontier and the closer an industry gets to the best frontier, the more the
industry has been successful in ‘catching-up’ and this is due to better use of technology
and equipment. How much the best frontier shifts at each industry’s observed input mix is
termed as technical change and this is due to the use of better technology and equipment.
The frontier version of productivity change consists of the ‘catching up’ and technical
change experienced by the industries. One caveat arising from this industry-based study is
that capital-intensive industries are compared to labour-intensive industries and this may
lead to biasedness in productivity indices. But little can be done about this since firm-level
data is yet to be made available for the Malaysian manufacturing sector.
The DEA methodology was initiated by Charnes et al. (1978) who built on the frontier
concept pioneered by Farrell (1957). DEA which uses linear programming techniques to
calculate the Malmquist index of TFP growth is chosen for the following reasons. First,
DEA is a major improvement over the translog index approach used by previous studies. As
shown by Fried et al. (1993), the translog index approach is flawed since it ignores technical
inefficiency and only calculates technical change, which is inaccurately interpreted as TFP
growth. But in the productivity literature, TFP growth is shown to be composed of both
technical change (frontier shift) and technical efficiency (catching up effect). Furthermore,
the translog index approach calculates TFP growth as a residual measuring ‘anything and
everything’ of output growth not accounted for by input growth. Unlike the residual ap-
proach, this exercise using DEA is important as it is able to identify the sources of TFP
growth and this enables better policy prescriptions since they can be made rather specific
to the particular source of low TFP growth.
Second, the DEA has some advantages over the stochastic frontier approach which calcu-
lates both technical efficiency and technical change components of TFP growth. Studies such
as Färe et al. (1989), Chaves and Cox (1990), and Callan (1991) show that nonparametric
techniques outperform parametric techniques in some situations. One important advantage
is that DEA envelopes observed input-output data without requiring a priori specification of
functional forms. Different specifications of the production function under the parametric
approach provide different results and this is a serious methodological problem. In addition,
Felipe (1999) shows how the modeling of the production function is a concept with little
theoretical justification as it simply represents a value added accounting identity.
Another advantage is that the nonparametric nature of DEA allows it to concentrate
on revealed best-practice frontiers rather than on central-tendency properties of frontiers.
Furthermore, as argued in Gong and Sickles (1992), DEA is more appealing than the
econometric model as inefficiency is likely to be correlated with the inputs. Lastly, DEA is
able to provide information on scale efficiency without the need for price data. However,

Das könnte Ihnen auch gefallen