-
World Bank Reprint Series: Number 394 R P 9
Heba Handoussa, Mieko Nishimizu, and John M. Page, Jr.
Pro ductivity Change mnEgyptian Public SedtorIndustries after
'The Op ninwg',1973-1979
Reprinted with permission from the Journal of Development
Economics, ol. 20 (1986), pp. 53-73,published by Elsevier Science
B.V., (North-Holland), Amsterdarn.
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
-
i
-
Journal of Development Economics 20 (1986) 53-73.
North-Holland
PRODUCTIVITY CHANGE IN EGYPTIAN PUBLIC SECTORINDUSTRIES AFTER
'THE OPENING', 1973-1979*
Heba HANDOUSSAAmerican University in Cairo, Cairo, Egypt
Mieko NISHIMIZU and John M. PAGE, Jr.The World Bank, Washington,
DC 20433, USA
Received January 1984, final version received July 1984
Policy reforms to accelerate growth and improve production
efficiency have been undertaken inEgypt since 1973. This paper
evaluates the impact of these reforms on Egypt's public sectorfirms
in terms of their productivity performance. We find an important
asymmetry in theconsequences of the reforms between the rapidly
expanding import substitution sector with highproductivity growth
and the stagnant traditional export sector. Our results also
indicate thatmuch of the observed productivity growth of the import
substitution sector may be ascribed toimprovements in capacity
utilization, and suggests that Egypt's industrial policies be
reassessedsince such a path may not be sustainable.
1. Introduction
Productivity change and the technical efficiency of industry are
one of themajor issues confronting the Egyptian economy. Mabro and
Radwan (1976)estimated a range of total factor productivity (TFP)
growth rates underalternative data and methodologies for Egyptian
manufacturing in thefollowing four periods:
1939-45 4.44- 6.70% per year1954-54 1.36- 3.35% per year1954-62
2.66- 2.88% per year1963/4-69/70 -1.90 - -2.20% per year
*This paper describes some results from a World Bank research
project: Productivity Changein Infant Industries, and was also
prepared as an input into a joint study between the ArabRepublic of
Egypt, Ministry of Industry and Mineral Wealth, and the World Bank
on 'TradeStrategy and Comparative Advantage in Egyptian Industry'.
We thank Kemal Dervis,Mahmoud Hillal, and participants of seminars
at the Research Program in Development Studiesat Princeton
University and at the University of Sussex for helpful comments on
earlier drafts ofthis paper. We thank Deborah Bateman for her
excellent research assistance. The views andinterpretations in this
paper are those of the authors and should not be attributed to the
WorldBank, to its affiliated organizations, or to any individual
acting on their behalf.
0304-3878/86/$3.50 © 1986, Elsevier Science Publishers B.V.
(North-Holland)
-
54 H. Handoussa et al., Productivity change in Egyptian public
sector industries
These estimates show a substantial productivity slowdown in
Egyptian
manufacturing since the interwar period. The three postwar
periods distin-
guished by Mabro and Radwan correspond to three policy regimes
that
significantly altered the production environment of Egyptian
industry.The period 1945-54 can best be characterized as an import
substitution
regime. The government pursued an 'Egyptianization policy'
directed at
creating a market economy with predominantly private Egyptian
ownership
(although the Egyptianization policy was reversed after the 1952
Revolution
in a brief effort to encourage foreign investment). In 1952, the
public sector's
share in GDP was about 13 percent, and in gross fixed capital
formation
(mostly in infrastructure) about 28 percent. The period 1954-62
was a transi-
tion period from a market to a centrally planned economy.
Government
intervention in productive activities steadily increased, and
nationalization
of foreign interests and major Egyptian financial concerns took
place. By
1960, the public sector's share in gross fixed capital formation
had risen to
about 74 percent. In both of these periods, TFP changes remained
positive
and were comparable to the average productivity performance of
other
countries.The transition was completed by the 1961 'Socialist
Revolution', when a
wholesale nationalization of industrial and financial
enterprises took place.
The period 1962-70 corresponds roughly to the central planning
regime
under 'Arab socialism'. Industrial activities in the private
sector were heavily
curtailed, and annual investment allocation plans and price
controls became
major instruments for the state control of the Egyptian economy.
Mabro and
Radwan's estimates show negative TFP change during this period,
repre-
senting a sharp break from the preceding twenty years.Shortly
after the October 1973 war, initiatives were taken to redirect
economic policies in Egypt. Partly in response to indications of
declining
industrial performance, the policy reforms collectively known as
'El-Infitah',
-r 'The Opening', were undertaken beginning in 1973. The basic
economic
objective of The Opening was to accelerate economic growth and
improve
production efficiency. Broadly interpreted, The Opening is
founded on three
sets of policy reforms. First, a series of new trade and
exchange control
regulations were brought about to gradually liberalize trade and
access to
foreign exchange. Second, a policy of gradual decentralization
in the control
of public sector firms was introduced to give greater autonomy
to public
sector managers. Third, a set of new investment laws were
formulated to
encourage foreign and domestic (both public and private)
producers to
introduce new products and technologies through joint-ventures,
and to
stimulate export activities.' These policies, taken together,
were intended
'The Opening has been conventionally associated only with this
third policy reform. We take
in this paper a broader interpretation of reformns under The
Opening.
-
H. Handoussa et al., Productivity change in Egyptian public
sector industries 55
gradually to shift the economy from a centrally planned system
to a mixedmarket economy.
Those reforms were not the only aspect of the post-1973
transformation ofthe Egyptian economy. The years 1973-79 were also
a period of rapidgrowth. Real GDP grew at about 8 to 9 percent per
year, due mostly tosubstantial increases in the external resources
available to the economythrough oil revenues, Suez Canal earnings,
Egyptian workers remittances,tourism, direct foreign investment,
and foreign aid. In particular, imple-mentation of policy reforms
on trade and exchange control liberalizationwas made easier by the
increasing availability of external resources. Theliberalized trade
regime removed supply-side constraints on imported inputsthat
plagued many firms under the preceding regime.
The years 1973-79 can therefore be characterized as the period
duringwhich both supply-side and demand-side constraints were
significantly re-laxed for public sector firms. The objective of
this paper is to assess theimpact of this confluence of policy
reforms and rapid growth on theproductivity performance of Egypt's
public sector firms. In particular, weshall investigate an
important asymmetry in the way aspects of The Openingaffected two
different groups of Egypt's public sector firms: firms engaged
inimport substitution activities, and firms who have exported
traditionally.
This paper applies a methodology recently developed by Nishimizu
andPage (1982), which permits the decomposition on TFP change into
'bestpractice' TFP change and technical efficiency change. The
methodology issummarized in section 2, where we also discuss
specification and estima-tion of the frontier production function
on which the productivity estimatesare based.2 Section 3 discusses
the empirical results, and conclusions arepresented in section
4.
2. The framework of analysis
As conventionally defined the production function describes a
productionprocess that is technically efficient, in that it is
impossible to do any better atthe given level of technological
knowledge. Technological change pushes theproduction function
outward, and improves the productivity of factor inputs.If, on the
other hand, one is technically inefficient and producing within
thetechnically feasible limits, changes in technical efficiency
also alter theproductivity of factor inputs. Technological change
always implies produc-tivity change. Productivity change, however,
does not necessarily imply tech-nological change.
The empirical estimate of the production function which
corresponds mostclosely with this definition is the frontier or
best practice production
2 For further details, see Nishirnizu and Page (1982).
-
56 H. Handoussa et al., Productivity change in Egyptian public
sector industries
function, which defines over any set of observations the maximum
output
obtained from a vector of measured inputs. It, thus, provides an
efficiency
standard which is based on the best practice actually observed
in a given
economic environment. This empirical concept of the production
frontier
may differ in application from the production function in
theory. The
estimated frontier will lie below the 'true' production
function, if, for
example, the observed best practice firms defining it are not
entirely free of
some form of technical inefficiency. For this reason, we refer
to the rate of
shifts of the frontier as best practice TFP change rather than
technological
change.We can represent the production relationship as
x(s, t) =g[z(s, t); s, t](1)
-
H. Handoussa et al., Productivity change in Egyptian public
sector industries 57
Eq. (3) represents a decomposition of actual TFP change into
threecomponents: best practice TFP change, technical efficiency
change, andoutput elasticity differences between the frontier and
the interior. The rate ofshift of the frontier, g(z, s, t),
represents the rate of best practice TFP change.Its movement over
time, however, must not be confused with changes inefficiency
relative to best practice. These effects are captured by e(s, t),
whichrepresents the rate at which any observed firm is moving
toward (e>O) oraway (e
-
58 H. Handoussa et al., Productivity change in Egyptian public
sector industries
deterministic, probablistic, and stochastic. We choose in this
paper to takethe deterministic approach, which permits us to
utilize the entire sample of
observations, and specify a one-sided error structure to
restrict all observedpoints in output space to lie on or below the
frontier. We therefore use alinear programming estimation
procedure, and minimize the sum of devia-tions from the frontier
subject to the on-or-below the frontier constraint.
4
We minimize the objective function which is linear in the
unknown
parameters,
T S
min E E [o + at + ± ,S,t2] + m + Imtt) In Zm(S, t)E sl ( m
t)
+ 2 flE , In Zm(S, t) In zJ(s, t) - In x(s, t) (5)m n
subject to the constraints of the model,
[cta + a,~t + 2f,2 ] + E (am + 3m,t) In Zm(S, t)m
+'Y 2 E .mn zm(s, t) In z,(s, t) > In x(s, t), s =1, ... , S,
t =1, . .. ,T7m n (6)
We also impose on the frontier the following set of restrictions
for constant
returns to scale:
XGOm= , Ypm.=O and Zfim t =O. (7)m m m
Finally, we impose concavity and monotonicity restrictions,
since the trans-
log form is neither monotonic nor concave for any arbitrary set
of argu-
ments. For concavity, we restrict #3mm to be non-positive. For
monotonicity,
4 The deterministic technique is the closest representation of
the theoretical concept of the
frontier as an outer boundary of the production set. It,
however, is more sensitive to errors inobservations compared with
the other two techniques which attempt to reduce the sensitivity
to'truly' random errors distinguished from efficiency distribution.
If one has sufficient informationto specify the random distribution
and efficiency distribution explicitly in the error structure,then
procedures have been suggested for the maximum likelihood
estimation of the frontier. Theliterature has yet to provide
guidance on the appropriate specification of the error
structure,however, and the statistical properties of the maximum
likelihood frontier estimators are notalways straightforward. The
deterministic technique, on the other hand, does have the
drawbackof presuming that all errors are due to the efficiency
distribution. For a recent comprehensivesurvey of the literature
and other important contributions on the formulation and estimation
of
frontier functions, see Aigner and Schmidt (1980).
-
H. Handoussa et al., Productivity change in Egyptian public
sector industries 59
we restrict am and a, to be non-negative, and
am + f3mtt + Efl,mn In z,(S, t) _ 0,(8)
(Xt + |ttt + Y. Bmt In zm(s, t) 0.m
We estimate the frontier using the geometric mean of the sample
as the pointof expansion for the translog approximation. The
estimated translog frontierparameters for each industry are
presented in table A.1 of the appendix.
3. The empirical results
Estimates of average annual rates of actual TFP change, best
practice TFPchange, and technical efficiency change are presented
for the 'average' firm(with mean level of output and inputs) in
each industry in table L.' There arethree important aspects to the
results.. First, many industries show highaverage rates of best
practice TFP change. Ten industries have best practiceTFP growth of
more than 2 percent per year, and only one industry
(Cottonproducts) has a stationary frontier. Second, the average gap
between actualand best practice TFP performance is widening over
time in most of theindustries. All but three industries show
negative rates of technical efficiencychange. Third, despite this
widening gap, actual TFP growth is stillrespectable for many of the
industries. Negative technical efficiency changesare smaller in
their absolute magnitudes than corresponding best practiceTFP
changes in all but five industries. Average firms in these
industries areexperiencing productivity improvements, although at
slower rates than theirpotential as defined by best practice.6
Table 1 shows seven of the fifteenindustries with actual TFP growth
of more than 2 percent per year.
5Thus, our estimates of actual TFP change are not adjusted for
the effect due to differencebetween observed interior output
elasticities and estimated frontier elasticities. We evaluate
bestpractice TFP change by combining input levels with the
estimated frontier parameters in eachyear, and taking simple
averages of consecutive time periods. Technical efficiency levels
ofaverage firms as defined in (2) can be obtained as the antilog of
the slack variables in our linearprogramming constraints (6). The
rates of technical efficiency change can be obtained by takingthe
log differences of this ratio in successive time periods. Actual
TFP change can be computedas the sum of the best practice TFP
change and technical efficiency change.6A portion of the discussion
of the results which follow focuses on differences between
bestpractice and interior firms. As noted in footnote 4, the
deterministic estimation techniqueascribes all deviations from best
practice to technical inefficiency. There is therefore a dangerwith
this technique that purely random components may give rise to
spurious identification ofbest practice firms and bias the
interpretation of the results. In the Egyptian case this risk
ispresent but not large. The fifteen individual sectors on which
the aggregate results are basedshow very similar patterns of
productivity differentials between best practice and interior
firms.Moreover, extensive investigations at the firm level, using
direct interviews, confirmed that in thegreat majority of cases
underlying technical and behavioral evidence could be found to
supportthe stratification of the sample into best practice and
average practice firms using the estimatedproduction frontiers.
These two arguments, taken together, lend substantial weight to the
use ofthe deterministic production function to establish best
practice in the Egyptian context andsupport the utility of our
analysis of the sources of productivity differentials between best
andaverage practice firms.
-
Table 1
Actual TFP change, best practice TFP change, and technical
efficiency change (in percent per year).'
Egypt (1973-79) Actual TFP change
Best-practice TechnicalTFP efficiency Actual TFP Japan USA US
and Japanese
Egyptian industries change change change (1955-73) (1947-73)
industries
FoodEdible oils 9.45 -1.84 7.61 221 0.01 FoodBeverages and
tobacco 2.21 3.74 5.95 0.28 TobaccoOther food 2.50 0.97 3.48
TextilesCotton products 0.00 -2.00 -2.00 1.70 1.54 Textiles mill
products
Other textiles 2.96 -1.56 1.40 1.94 0.67 Apparel
ChemicalsPaper 7.71 -5.36 2.34 1.62 0.01 PaperBasic chemicals
0.91 -3.74 -2.83 2.50 2.33 ChemicalsFertilizer 1.66 11.51
13.17Rubber and plastic 1.92 -3.49 -1.57 -1.22 0.99 Rubber and
plastic a
Light consumer goods 4.43 -5.75 -1.32
Metal products and engineeringIron and steel 3.16 -2.51 0.65
0.96 -0.49 Iron and steel
Transportation equipment 8.18 -3.66 4.52 2.53 0.74
Transportation equipment
Fabricated metals and 0.84 0.19 Fabricated metals
machinery 5.68 -5.22 0.46 3.14 0.55 MachineryElectrical
machinery 9.03 -5.22 3.81 4.42 1.48 Electrical machinery
China and glass 0.31 -0.46 -0.15 1.73 0.23 Stone, clay and
glass
'Source: The estimates for Japan are from Nishimizu and Robinson
(1983). The estimates for USA are from Gollop and Jorgenson
(1980).
-
I I
H. Handoussa et al., Productivity change in Egyptian public
sector industries 61
TFP growth in mature industrialized economies over a
sufficiently longperiod of time is conventionally believed to
approximate rates of techno-logical change. The stylized facts
accumulated in the productivity literatureshow that one can expect
these long-run rates to cluster closely around 1 to2 percent per
year, and this is observed in the productivity growth estimatesfor
roughly comparable industries in Japan and the USA given in table
1.TFP growth for developing countries, on the other hand, is often
influencedsignificantly by factors other than technological change
especially whenimportant changes occur in demand conditions,
production constraints, andtrade and industrial policies.7 TFP
estimates for developing countries tend tobe distributed more
widely around 1 to 2 percent as we observe in Egypt. Itis
reasonable to consider, therefore, particularly in light of the
high rates ofestimated best practice TFP change, that not only
actual but also bestpractice TFP change in Egypt reflect short-run
adjustment by firms to thenew production environment. 8 Best
practice firms may be those which weremost successful in adjusting
rapidly.
The period 1973-79 in Egypt cannot in general be characterized
as an eraduring which new industrial activities, new plants, or new
firms came onstream to add to the productive capacity of the public
sector. Rather, mostof the industrial production capacity was in
place before 1973. This industrialcapacity, however, was reportedly
severely underutilized before 1973. The lowrate of capacity
utilization was in part due to stagnant domestic demand, butwas
mostly caused by acute shortages of imported material inputs.
Firmsengaged in import substituting activities were particularly
hard hit, sincetheir import requirements were high and demand was
slow growing. Importrequirements of traditional export activities
in Egypt were low, and they hadthe secure bilateral trade markets
with the USSR and other Eastern Bloccountries. 9
By 1973 centralized control of industry and the combination of
supply-sideconstraints on imported inputs with slow growth in
aggregate demand hadresulted in low productivity levels in most
branches of industry and in littledispersion among individual firms
in relative levels of TFP. It is notsurprising, therefore, that the
single most important policy change thataffected public sector
firms was the liberalization of trade and foreignexchange control.
The liberalization, however, was asymmetric in that it
'See Nishimizu and Robinson (1983), for example.8 This is
precisely the reason why we chose not to refer to the shift of the
frontier production
function as technological change, since part of the estimated
shift may not represent reduction inlong-run average cost.
9 The most important traditional export activity is Cotton
products. In addition, there arefirms from each of the following
industries: Other textiles, Food products, and Light consumergoods.
All other activities can be considered import substituting
activities. Estimates of effectiverates of protection at the
sectoral level are contained in Handoussa (1980) and Hansen
andNashashibi (1975).
I.D.E. C
-
62 H. Handoussa et al.. Productivity change in Egyptian public
sector industries
permitted easier access to imported inputs but continued to
protect domesticmarkets for the output of public sector firms from
import competition. Themajor constraint on capacity utilization of
import substituting firms was thusremoved, aided greatly by rapidly
increasing domestic demand. Short-runadjustment to changing supply-
and demand-side constraints created anenvironment in which import
substituting firms could and did react withvarying speed to the new
economic opportunities. These varying rates ofadjustment resulted
in a widening gap between best practice and averageTFP levels
within import substituting industries and in some increase in
thevariance of TFP levels across firms. Since the comparative
advantage ofEgypt's traditional export firms was natural resource
based, liberalization onthe input side had minimal impact on them.
The impact of the new traderegime, however, did not remain neutral
for many of the traditional exportfirms, since with The Opening
bilateral trade markets closed and firms wereforced to look for new
markets in convertible currency areas.
The policy reforms of The Opening and the rapid increase in
aggregatedemand following 1973 thus benefitted import substituting
firms but not thetraditional export firms. Trend output and TFP
(actual) growth rates forimport substituting firms and for
traditional export firms are presented intable 2. The differences
in output growth between the two groups arestriking. The overall
trend growth rate in output for the entire sample ofpublic sector
firms is 8.9 percent. For import substituting firms the trendoutput
growth rate is 11 percent, while for traditional export firms the
trend
Table 2
Trend growth rates of output and total factor productivity,
Egypt's industrial sector,1973-79.'
Output growth TFP growth
Best BestAll Interior practice All Interior practicefirms firms
firms firms firms firms
Import substitutingfirms I1.00b 11.96 10.08 2.40' 2.64 2.16
Traditional exportfirms 0.00b 0.00 0.00 - 1.91 -3. 4 6 d 0 .0 0
d
All firms 8.94 9.06 9.19 1.17 0.00 2.02
'All non-zero trend growth rates are significant at the 0.01
level. TFP growth rates arethe actual TFP rates.
bImport substituting and traditional export firms differ at the
0.10 level in the timetrend regression with a dummy variable for
the latter.
'lmport substituting and traditional export firms differ at the
0.01 level in the timetrend regression with a dummy variable for
the latter.
dinterior and best practice firms differ at the 0.01 level in
the time trend regression witha dummy variable for the latter.
-
H. Handoussa et al., Productivity change in Egyptian public
sector industries 63
growth rate is not significantly different from zero. The
difference in growthrates between import substituting firms and
traditional export firms issignificant (at the 0.10 level). In each
group, the output growth rates for firmsdefining best practice do
not differ significantly from those of interior firms.Thus, the
rapid increase in industrial output was almost wholly confined
toimport substituting firms. Among traditional export firms
increases indomestic demand were offset by reductions in deliveries
to foreign buyers,and there was no overall trend in output
growth.
Trend TFP growth rates presented in table 2 further reinforce
theperception of dramatic differences between the two groups. The
trend growthrate of TFP for the public sector as a whole is about
1.2 percent. In importsubstituting firms the trend growth rate is
2.4 percent, while for traditionalexport firms it is - 1.9 percent.
The growth rates differ significantly betweenthe two groups (at the
0.01 level). Among import substituting firms there isno
statistically significant difference between the trend rates of TFP
growthof best practice and of interior firms. Among traditional
export firms,however, the trend rate of TFP growth of interior
firms is about -3.5percent and significantly lower than that of
best practice firms (at the 0.01level).
Taken together, these results suggest that there may exist a
strongassociation between output growth and productivity
performance in theEgyptian public sector. Fast growing, import
substituting firms experiencedrates of productivity change which
greatly exceeded those exhibited bytraditional exporters.
The important question to ask is what underlies the observed
asymmetryin the productivity performance between the import
substituting firms andtraditional export firms. The discussion so
far suggests capacity utilization asthe key differentiating factor.
Rapid improvement in capacity utilization fromlow pre-1973 levels
towards optimum levels may underlie the high rates ofmeasured TFP
change in import substituting firms (both best practice andinterior
firms). Traditional export firms on the other hand, probably failed
toachieve significant changes in capacity utilization. Output
growth was low ornegative, and most firms encountered difficulty in
shifting export markets asthe opportunities for trade with
non-convertible currency areas diminished.'0
The capacity utilization adjustment problems arise because there
are quasi-fixed inputs in the short run which can only be changed
at increasingmarginal cost of adjustment. Such quasi-fixedness is
usually associated with
'°Ikram (1980, p. 240) reaches a similar conclusion.
Unfortunately, reliable and independentestimates of capacity
utilization rates do not exist for the period. Nevertheless, we are
reasonablyconfident of our interpretation which is largely based on
plant visits and interviews with manyof the sample firms.
-
64 H. Handoussa et al., Productivity change in Egyptian public
sector industries
capital stocks." Strictly speaking, our analytical framework
presumes long-
run production equilibrium at the frontier. If this is true, the
estimated shift
of the frontier is equivalent to the shift in long-run average
cost. If not, part
of the estimated shift is the movement along short-run average
cost schedule
as the capacity utilization changes over time.'2
If firms are operating at optimum capacity utilization,
variations in output
growth should mostly be accounted for by variations in input
growth; and
there should be little systematic correlation between measured
(actual) TFP
growth and input growth or output growth. If, on the other hand,
firms are
increasing capacity utilization towards optimal rates,
variations in output
growth among firms and variations in their input growth should
have
weaker correlation. Although there should be no systematic
correlation
between TFP growth and input growth, there should be a
substantial
correlation between TFP growth and output growth.
This argument can be explained in terms of the following
three
relationships:
(A) output growth = a1 + b, input growth,"3
(B) TFP growth = a2 + b2 input growth,
(C) TFP growth = a3 + b3 output growth.
There should be high correlation in (A), and since output growth
is always
the sum of TFP growth and input growth, a1 should equal the
average rate
of TFP growth and b, should equal unity. There should be low
correlation
in (B) or (C), and b2 and b3 should be zero.'4
Since our measure of TFP change for each firm is derived by
applying
output elasticities of each factor input evaluated at the
estimated frontier
production function, b1 in (A) should be unity for the best
practice firms who
define the frontier regardless of the capacity utilization
factor in their TFP
change. For interior firms, however, this need not be the case.
If a quasi-fixed
factor is more underutilized in interior firms relative to best
practice firms, its
"In the case of Egypt's public sector firms, however, it is
probably appropriate to consider
labor as quasi-fixed as well. This is the consequence of
employment policies which not only
make it difficult, if not impossible, to lay off workers, but
also require compulsory addition of
workers to public sector firms in labor categories whose tasks
cannot be easily changed. See
Berndt and Fuss (1981).'2 See Berndt and Fuss (1981)."3 By input
growth, we mean total factor input growth which is the weighted
average (by
output elasticities) of capital, labor, and material input
growth rates.1 4 Given that output growth is the sum of TFP growth
and input growth, a2 = a, and b2 = b, -1.
A similar approach was adopted by Bruton (1967) in analyzing
capacity utilization and TFP
change in Latin American countries.
-
H. Handoussa et al., Productivity change in Egyptian public
sector industries 65
marginal product should be less than that implied at the
frontier. Bydefinition, quasi-fixed factor must be growing
relatively slowly comparedwith other inputs. It is likely, that we
may have overvalued the contributionof inputs which are growing
relatively slowly, and consequently undervaluedthe contribution of
those inputs which are growing relatively more rapidlyfor interior
firms. Thus, in relationship (A), b1 should be significantly
greaterthan unity for the interior firms, and a, should be
significantly less than theaverage rate of measured TFP change."5
If capacity utilization is animportant component of measured TFP
change, we should expect b3 to besignificantly positive for both
best practice and interior firms. The greater theforce of capacity
utilization in measured TFP change, the greater should bethe value
of b,. Since a3 can be interpreted as the average rate of TFPchange
net of the capacity utilization component, the greater the force
ofcapacity utilization, the lower should be the value of a3.
We have estimated these simple relationships (A), (B) and (C)
usingordinary least squares, on the average annual growth rates of
output, input,and actual TFP for 96 firms, separated into two
groups: (1) 67 firms engagedin import substituting activities, and
(2) 29 firms engaged in traditionalexport activities. In each
relationship for each group, we enter intercept andslope dummy
variables for best practice firms.' 6 The regression results
arepresented in table 3.
The most striking aspect of the regression results is the
contrast betweenthe import substituting firms and traditional
export firms in the relationship(C). For the import substituting
firms, 75 percent of the variance in TFPgrowth can be attributed to
output growth. For the traditional export firms,the multiple
correlation coefficient is much lower at 24 percent. Theestimated
'elasticities' of TFP growth with respect to output growth (b3)
aresignificantly positive (at the 0.01 level) for both best
practice and interiorfirms in the import substituting group. For
the exporting group, however,they are not significantly different
from zero (at the 0.05 level). The elasticityis significantly
higher for import substituting interior firms compared to
bestpractice firms: TFP growth improves by 0.64 percent for a one
percentincrease in output growth for the interior firms, and by
0.42 percent for thebest practice firms. The estimated intercepts
show that TFP growth forimport substituting firms is significantly
negative when output growth is zero:as much as - 3.6 percent for
the interior firms, and - 1.4 percent for the bestpractice
firms.
"5 See Berndt and Fuss (1981). They argue that the correct
adjustment for capacity utilizationis in terms of marginal products
of quasi-fixed inputs, and not in terms of the quantity of
theseinputs which has been the traditional method.
'6Applying covariance analyses to our panel data indicates that
there are statisticallysignificant differences in the estimated
regressions between the import substituting firms andtraditional
export firms. We therefore report separate regressions for each
group. Although thebest practice firms are not always different
from the interior firms (particularly in exportinggroup), we report
the regressions with dummy variables for symmetry.
-
ol
Table 3TFP change and capacity utilization in Egypt's public
sector firms (1973-79).' -
Import substituting firms (67 firms) R2 Traditional export firms
(29 firms) R22
(A) output growth= -0.018+0.034F+(1.700-0.542F) inpul growh 0.63
output growth -0.003+0.015F+ (0.667+0.359F) input growth 08(0.020)
(0.026) (0.210) (0.277) (0.012) (0.021) (0.150) (0.319)
(B) TFP growth= 0.018 +0.034F+(0.700-0542 F) input growth 0.16
TFP growth= -0.003+0.015F+(-0.332+0.359F) input growth 0.25
>(0.020) (0.026) (0.210) (0.277) (0.012)(0.021) (0.150) (0.319)
'
TFP growth= -0.029+0.034F+ (0.309-0 148F) output growth (.'4
z(C) TFP growth --0.036+0.022F+(0.644-0.221F) output growth 0.75
(0.010) (0.022) (0.154) (( 299) >
(0009) (0.013) (0.052) (0.087)
'Standard error is presented in parentheses under each estimated
coefficient. F is the dummy variable equal to unity for best
practice firms and zerootherwise.
'5'5.
-
I~~~~~~~~~~ I
H. Handoussa et al., Productivity change in Egyptian public
sector industries 67
These results are consistent with our interpretation that
improvements incapacity utilization may underlie measured TFP
growth in import substi-tuting firms while this was not the case
for exporting firms. Furthermore, thecapacity utilization factor in
measured TFP growth seems to be significantlymore important for
interior firms as opposed to best practice firms amongthe import
substituting firms. This is also consistent with our
interpretationthat the best practice firms are those who were most
successful in adjustingrapidly to the new production environment
after The Opening.
The estimated coefficients of relationship (A) for the import
substitutingfirms are also consistent with this interpretation. The
coefficient for inputgrowth (b,) is significantly greater than
unity (at the 0.01 level) for interiorfirms, indicating that we
have underestimated the input contribution tooutput growth for
these firms by applying the frontier output elasticities.' 7
Itimplies, in turn, that measured TFP change adjusted for capacity
utilizationshould be lower. The estimated intercept for both
interior and best practicefirms is not significantly different from
zero, implying that, on average, all ofthe measured TFP change
could be attributed to improvements in capacityutilization.
The estimated coefficients for relationship (A) invite a
different interpret-ation for traditional export firms. The
coefficient for input growth (b,) issignificantly less than unity
for interior firms, indicating that we haveoverestimated the input
contribution by using frontier elasticities, andtherefore
underestimated TFP change. (The estimated intercept is notdifferent
from zero for both interior firms and best practice firms.) This
resultis probably due to a phenomenon unique to the cotton textile
industry,which forms the bulk of our sample of traditional export
firms. After TheOpening, many firms in the industry were forced to
shift away from bilateralexport markets in the Eastern Bloc to
convertible currency markets and tothe domestic market. The loss of
the bilateral markets meant a shift from thehigher negotiated
prices in the Eastern Bloc to lower prices in the
convertiblecurrency areas. In the domestic market, many firms were
subject to increasesin the proportion of products assigned by the
government to be sold at lowcontrolled prices. In both cases, unit
values for output declined without anaccompanying decline in the
quality of the product. The best practice firmsare those who were
least subjected to these changes and/or those whoadjusted product
mix and cost structure quickly to new prices. The interiorfirms are
those who were substantially affected by the changes and
haddifficulty in adjusting particularly in the face of the dual
price and quantitycontrols imposed by the government. The
constraint in the adjustmentprocess was the difficulty in 'scaling
down' quasi-fixed inputs. By using the
"7 The estimated b, for the best practice firms is not different
from unity as expected. This isalso the case for the traditional
export firms.
-
68 H. Handoussa et al., Productivity change in Egyptian public
sector industries
frontier elasticities, therefore, we have assigned lower weights
to quasi-fixed
inputs and consequently higher weights to other more rapidly
growing
inputs. The overall bias on estimated input growth is upward,
which is
consistent with the result that the coefficient b1 is
significantly less than
unity. As we have stated earlier, this implies in turn that our
measured TFP
change for interior traditional export firms is probably biased
downward.These results are necessary but not sufficient
confirmation of the hypo-
thesis that capacity utilization determined productivity change
in Egypt,
since we can always question the causality relationship of the
estimated
regression equations. In particular, one can always interpret
the relationship
(C) to mean that rapid TFP growth is correlated with high growth
in output
since the former reduces unit cost of production. Falling unit
costs may
engender rapid price reductions leading to greater demand.
4. Conclusions
The main objective of this paper was to assess the impact of The
Opening
on the performance of Egypt's public sector industry in the
period of rapid
growth following 1973. In particular, we noted and investigated
an important
asymmetry in the way policy reforms of The Opening favored firms
engaged
in import substituting activities, perhaps at the expense of
firms engaged in
traditional export activities.Given this objective, we have
focused on one measure of industrial
performance, total factor productivity. Our results indicated
that Egypt'spublic sector industries experienced impressive rates
of TFP growth by
international standards. Our decomposition of TFP change into
best practice
TFP change and change in technical efficiency revealed that in
most
industries the rates of best practice TFP growth were very high
relative to
what is usually expected as the long term rates of technological
change.
These high rates of best practice TFP growth were offset
somewhat by
deteriorating technical efficiency. In most industries in Egypt
the average
firm was experiencing a widening gap between its level of TFP
and the level
of best practice. This pattern of TFP change is consistent with
the view that
public sector firms adjusted to the liberalization of the trade
regime, increasesin effective demand, and increased autonomy at
different rates. Best practicefirms were those which were most able
to adapt quickly to the changing
production environment while other firms adjusted more slowly.We
view the rapid overall rates of productivity change as indicative
of the
sucess of the liberalization efforts undertaken following 1973.
The deterior-
ation of the average level of technical efficiency and the
increased dispersionof relative levels of efficiency which occurred
in most industries is probablyan inevitable consequence of the
movement from highly centralized direction
of public sector firms to a more decentralized environment for
production.
-
I I
H. Handoussa et al., Productivity change in Egyptian public
sector industries 69
It is important to remember, however, that we are viewing a
relativelyshort run process of adjustment. Here, we noted the
asymmetry in theliberalization efforts of The Opening, and the
great difference in theproduction and productivity performance
between import substituting firmsand traditional export firms. High
rates of measured TFP change in Egypt'spublic sector basically
occurred in the import substituting firms whoseoutput expansion was
very rapid, while traditional export firms showed noimprovement or
deterioration in TFP and stagnant output growth. Theanalysis of
this difference led us to propose that much of the measured
TFPchange in Egypt's public sector may be due to improvements in
capacityutilization among import substituting firms. Although far
from being con-clusive, our simple empirical analysis supports such
a view.
If we are correct in our assessment that the improvement in
capacityutilization was the prime mover of TFP change, it may have
implications forindustrial policy in Egypt since such a path is not
sustainable. In the longerrun, efforts to adopt or generate
technical innovations will be necessary ifindustries in Egypt's
public sector are to continue to improve their produc-tive
efficiency, and if Egypt is to regain the sustained rate of
productivitychange characteristic of the period 1945-62.
Appendix
Data and assumptions
Data on gross output, material inputs, capital stock and
investment, andemployment by enterprise are provided by the annual
standardized accountsof Egyptian public sector enterprises
reporting to the Ministry of Industryand Mineral Wealth.
Producer price indices for output were constructed on a firm
specific basisby rebasing the value of output at previous years'
prices reported by eachfirm to constant 1972 prices. Material input
prices were constructed bycombining commodity specific price
indices for major inputs (with costshares greater than 5 percent),
the wholesale price index for all otherdomestic inputs, and an
import price index for all other imported inputs.Commodity specific
international prices for major imported inputs wereobtained from
the World Bank's Commodity Trade and Price Trends and TheBank of
Japan Price Indexes Annual, and adjusted for transportation
marginsand exchange rate changes. Commodity specific domestic
prices for majorinputs produced in Egypt were provided by the
Ministry of Industry. Thewholesale price index is published by
Central Authority for Public Mobiliz-ation and Statistics, and the
import price index was constructed by combin-ing the OECD export
price index with an exchange rate adjustment factor.
-
70 H. Handoussa et al., Productivity change in Egyptian public
sector industries
Capital stocks in constant prices were constructed by deflating
the annualvalue of net investment provided in the standardized
accounts by a capitalinvestment deflator reflecting price changes
in construction costs, machineryand equipment, and other assets and
adding these constant price investmentvalues to a base year (1972)
capital stock estimates. Weights for the capitalinvestment deflator
were derived from data on the composition of gross fixedinvestment
provided in World Bank (1980). The standardized accountsprovide
base year capital stock at historic cost in 1972 prices. We
generatereplacement cost base year capital stock estimates by
adjusting the historiccosts stocks with a perpetual inventory
estimate of the relationship betweenhistoric cost and replacement
cost valuations, given a rate of depreciation of0.085, inflation in
capital goods prices of 8 percent, and an estimated annualrate of
growth of the capital stock of 5 percent per annum between 1963
and1973.
Table A.lFrontier translog production function parameters, Egypt
1973-79.'
Food Textiles
Edible Beverages Other Cotton Otheroils and tobacco food
products textiles
a0 9.58418 10.72291 9.35827 9.80521 9.73027a, 0.09465 0.02220
0.02518 - 0.02954/3, 0.01209 -0.00051 - 0.00281 - - 0.00933aK
0.22930 - 0.26700 0.32342 0.13573arL 0.12716 0.02916 0.29492
0.35695 -aM 0.64355 0.97084 0.43808 0.31963 0.86427IJKK - -0.13360
-0.23713 -0.18838fALL -0.44330 - -0.32126 -0.08080 -I;MM -0.11253 -
- -0.45248 -0.18838/3RL 0.16539 0.22743 -0.06728 -/KM -0.16539 -
-0.09383 0.30441 0.18838/LM 0.27791 - 0.09383 0.14807 -PK, 0.07720
- -0.00437 - 0.00330/3LI -0.02215 0.00972 -0.04476 -PMl -0.05506
-0.00972 0.04913 - -0.00330
-
H. Handoussa et al., Productivity change in Egyptian public
sector industries 71
Table A.1 (continued)
Metal products and engineering
FabricatedIron and Transportation metals and Electrical China
andsteel equipment machinery machinery glass
CXO 9.42521 9.17384 8.31111 8.77156 8.11484C!, 0.03206 0.08246
0.05591 0.09115 0.00317fit, 0.00363 0.00287 -0.00963 0.01918
-0.00017a,K 0.17065 0.09290 0.25567 0.04185 0.26415at, 0.23175
0.29103 0.23923 0.22330 -am 0.59759 0.61607 0.50509 0.73485
0.73585&KK -0.02528 -0.07615 -0.03165 - -fiLL -0.19470 -0.30936
- -fiMM - -0.25229 -0.03165 -PKL 0.10999 0.06661 - -#KM -0.08471
0.00954 0.03165 -fiLM 0.08471 0.24275 - - -PK, 0.02575 -0.00070
0.00751 -0.01395 0.00479fLI -0.00405 -0.06961 -0.07974 0.02809 -fMl
-0.02170 0.07031 0.07224 -0.01414 --0.00479
Chemicals
LightBasic Rubber and consumer
Paper chemicals Fertilizer plastic goods
ao 9.05293 8.63793 8.77953 8.95663 7.91731a, 0.07320 0.00909
0.01667 0.01896 0.03999f,,, -0.01755 0.00303 0.00109 0.00145
0.00718OCK 0.24242 - 0.50197 0.05920 0.21334GCL 0.31865 0.29981 -
0.00255 0.40987aM 0.43893 0.70019 0.49803 0.93825 0.37679fKk
-0.21857 - -0.16975 - -0.08655fiLL -0.25610 -0.02466 - -
-0.54068fMM -0.28271 -0.02466 -0.16975 -fiKL 0.09598 - - -
0.31362&KM 0.12259 - 0.16975 _ --0.22707fiLM 0.16013 0.02466 -
- 0.22707fKi -0.01054 - 0.01902 -0.01973 -0.01176fLI -0.02232 -
-0.00085 --0.02014flK 0.03285 -0.01902 0.02058 0.03189
at= time, K = capital, L = labor, M =material.
-
72 H. Handoussa et al., Productivity change in Egyptian public
sector industries
References
Aigner, D.J. and S.F. Chu, 1968, On estimating the industry
production function, AmericanEconomic Review 58, no. 4,
826-839.
Aigner, D.J. and P. Schmidt, eds., 1980, Specification and
estimation of frontier production,profit and cost functions, Annals
of Applied Econometrics 1980- 2, suppl., Journal ofEconomics 13,
no. 1.
Aigner, D.J., C.A.K. Lovell and P. Schmidt, 1977, Formulation
and estimation of stochasticfrontier production function models,
Journal of Econometrics 6, no. 1, 21-37.
Berndt, E.R. and M.A. Fuss, 1981, Productivity measurement using
capital asset valuation toadjust for variation in utilization,
Paper presented at the Econometric Society SummerMeetings (San
Diego, CA).
Bruton, H.J., 1967, Productivity growth in Latin America,
American Economic Review LVII, no.5,1099-1116.
Caves, D.W. and L.R. Christensen, 1980, The relative efficiency
of public and private firms in acompetitive environment: The case
of Canadian railroads, Journal of Political Economy 88,no. 5,
958-976.
Caves, D.W., L.R. Christensen and W.E. Diewert, 1980,
Multilateral comparisons of output,input, and productivity using
superlative index numbers, The Economic Journal 92, no.
365,73-86.
Caves, D.W., L.R. Christensen and W.E. Diewert, 1981, The
economic theory of index numbersand the measurement of input,
output, and productivity, Econometrica 50, no. 6, 1393-1414.
Christensen, L.R., D.W. Jorgenson and L.J. Lau, 1971, Conjugate
duality and the transcendentallogarithmic production function,
Econometrica 39, no. 4, 255-256.
Christensen, L.R., D.W. Jorgenson and L.J. Lau, 1973,
Transcendental logarithmic productionfrontiers, Review of Economics
and Statistics 55, no. 1, 28-45.
Diewert, W.E., 1976, Exact and superlative index numbers,
Journal of Econometrics 4, no. 2,115-146.
Farrell, M.J., 1957, The measurement of productive efficiency,
Journal of the Royal StatisticalSociety, Series A 120, part III,
11-181.
Forsund, F.R. and L. Hjalmarsson, 1979, Frontier production
functions and technical progress,A study of general milk processing
in Swedish dairy plants, Econometrica 47, no. 4, 883-900.
Gollop and Jorgenson, 1980, U.S. Economic growth: 1948-1973,
Mimeo.Handoussa, H., 1980, The impact of economic liberalization on
the performance of Egypt's
public sector industry, Paper presented at Second B.A.P.E.G.
Conference by BostonUniversity and Harvard Institute for
International Development.
Hansen, B. and K. Nashashibi, 1975, Foreign trade regimes and
economic development: Egypt,A Special Conference Series on Foreign
Trade Regimes and Economic Development, Vol. IV(National Bureau of
Economic Research, New York).
Ikram, K., 1980, Egypt: Economic management in a period of
transition (The Johns HopkinsUniversity Press, Baltimore, MD).
Kopp, R.J., 1981, The measurement of productive efficiency: A
reconsideration, QuarterlyJournal of Economics 96, no. 3,
477-504.
Lau, L.J., 1978, Testing and imposing monotonicity, convexity,
and quasi-convexity constraints,in: M. Fuss and D. McFadden, eds.,
Production economics: A dual approach to theory andapplications,
Vol. 1 (North-Holland, Amsterdam).
Mabro, R. and S. Radwan, 1976, The industrialization of Egypt,
1939-1973 (Clarendon Press,Oxford).
Malmqvist, S., 1953, Index numbers and indifference surfaces,
Trabajos de Estatistica 4, 209-242.Meeusen, W. and J. van den
Broek, 1977, Efficiency estimation from Cobb-Douglas production
functions with composed error, International Economic Review 18,
no. 2, 435-444.Nishimizu, M. and J.M. Page, Jr., 1982, Total factor
productivity growth, technological progress
and technical efficiency change: Dimensions of productivity
change in Yugoslavia, 1965-1978,The Economic Journal 92, no. 368,
920-936.
Nishimizu, M. and S. Robinson, 1983, Sectoral productivity
growth: A comparative analysis offour countries, Discussion paper
no. 52 (Development Research Department, World Bank,Washington,
DC).
-
I I
H. Handoussa et al., Productivity change in Egyptian Public
sector industries 73
Schmidt, P., 1976, On the statistical estimation of parametric
frontier production functions,Review of Economics and Statistics
58, no. 2, 238-239.
Solow, R.M., 1957, Technical change and the aggregate production
function, Review ofEconomics and Statistics 39, no. 3, 312-320.
Timmer, C. Peter, 1970, On measuring technical efficiency, Food
Research Institute Studies 9,99-171.
Timmer, C. Peter, 1971, Using a probabilistic frontier
production function to measure technicalefficiency, Journal of
Political Economy 79, no. 4, 776-794.
World Bank, 1980, Egypt's past performance in real capital
accumulation 1974-1979 (Washing-ton, DC).
-
II
-
THE WORLD BANK
Headquarters1818 H Street, NW.Washington, D.C. 20433, U.S.A.
Telephone: (202) 477-1234Telex: WUI 64145 WORLDBANK
RCA 248423 WORLDBKCable address: INTBAFRAD
WASHINGTONDC
European Office66, avenue d'1ena75116 Paris, France
Telephone: (1) 47.23.54.21Telex: 842-620628
Tokyo OfficeKokusai Building1-1, Marunouchi 3-chomeChiyoda-ku,
Tokyo 100, Japan
Telephone: (03) 214-5001Telex: 781-26838
The full range of World Bank publications, both free and for
sale, is describedin the World Bank Catalog of Publications, and of
the continuing research programof the World Bank, in World Bank
Research Program: Abstracts of Current Studies.The most recent
edition of each is available without charge from:
PUBLICATIONS SALES UNITTHE WORLD BANK1818 H STREET,
N.W.WASHINGTON, D.C. 20433U.S.A.
ISSN 0253-2131