Wei J f
POLICY RESEARCH WORKING PAPER 28 92
Productivity or Endowments?
Sectoral Evidence for Hong Kong's Aggregate Growth
Hiau Looi Kee
The World Bank
Development Research Group
Trade
September 2002
Poi-icy REsEARCH WORKING PAPER 2892
Abstract
Kee provides sectoral evidence that sheds new light on the maniufacturing sector is also unimiipressive. The
the current debate regarding the sources of growth of the manufacturing sector is more labor intensive and its
East Asian miracle. The author tests both the growth is hindered by the reallocation of resources into
productivity-driven and endowment-driven hypotheses the services sector as a result of the growth of capital
using Hong Kong's sectoral data. The results show that endowments and imports. Overall, sectoral evidence
most of the growth in the services sector is driven by the supports the endowment-driveni hypotihesis for Hong
rapidly accumLulating capital endowments, and riot by Kong's aggregate growth.
productivity growth. ln addition, productivity growth in
This paper-a product of Trade, Development Research Group-is part of a larger effort in the group to study the
relationship between trade arid growth. Copies of the paper are availablc free from the World 1Bank, I818 H Street NW,
Washington, DC 20433. Please contact Pauliria Flewitt, room MC3-333, telephone 202-473-2724, fax 202-522-1159,
email address pflewitt@worldbank.org. Policy Research Working Papers are also posted on the Web at littp://
econ.worldbank.org. The author may be contacted at hikee@vworldbank.org. September 2002. (30 pages)
The Po/ic)y Research Working Paper Series disseminates the findings of wuork in progress to encourage the exchange of ideas about
development issues. Ant objective ofttbe series is to get the findintgs ouit qutickly. even if the presentations are less thani fully polished. The
papers carry the nanmes of the auithors and should be cited accordinzgly. The findizgs, interpretations, and conclnsio,zs expressed in this
paper are entirely those of the authors. They dto not necessarily represent the viewv of the W7orld Bank, its Execotive Directors, or the
countries the) represent.
P'roduced by the Research Advisory Staff
Productivity or Endowments?
Sectoral Evidence for Hong Kong's Aggregate Growth
Hiau Looi Keel
The World Bank
JEL #: F14, F43, 047, 053
Keywords: East Asia miracle; Endowments; Productivity; Rybczynski elasticity
i Development Research Group, The World Bank, 1818 H Street, N.W., Washington, DC 20433. Tel: (202) 473
4155; Fax: (202) 522 1159; E-mail: hlkeeOworldbank.org. I would like to give special thanks to Robert Feenstra for
his insightful guidance and comments. Feedback from Lee Branstetter, Deborah Swenson, Catherine Morrison Paul,
Kaoru Nabeshima, Mary Amiti, and Marcelo Olarreaga is gratefully acknowledged.
1. Introduction
The history-defying growth of the four East Asian newly industrialized economies (NIEs) in the
past three decades has fascinated economists and policy makers around the world. After more than
a decade of extensive research based on the aggregate statistics of the economies, the literature
has offered two hypotheses regarding the "economic miracle": the productivity-driven and the
endowments-driven hypotheses. To date, there is still an ongoing debate regarding which of the
two is the more important source of growth of these economies. The goal of this paper is to provide
consistent sectoral evidence that may substantiate or invalidate these aggregate findings, and to
shed new light on the debate.
The productivity-driven hypothesis originated from the new growth theory, which emphasizes
the role of productivity growth. Lucas (1988) introduces the effect of trade on productivity growth
through a learning-by-doing mechanism. He proposes that the growth of the East Asian NIEs is a
result of productivity growth, which in turn is due to the production experience accumulated in the
export markets. Subsequent papers by Young (1991) and Lucas (1993) explore the growth effects
of trade in a similar way. Thus, this school postulates that the growth of the four East Asian NIEs
is a result of productivity growth that is associated with export growth.
To provide a theory of sustainable long-run growth that is consistent with the empirical findings,
Findlay (1996) and Ventura (1997) formalize the endowment-driven hypothesis. Ventura shows
that in a general equilibrium setting, a small open economy can sustain high growth through the
Rybczynski effects of factor accumulation. Given that factor prices are equalized through the
free trading of goods, when an economy experiences growth in a factor, say capital, the capital-
intensive industries in the economy will grow at the expense of the noncapital-intensive industries.
Reallocation of resources across sectors makes it possible to defy diminishing returns to factor
accumulation as long as the economy is not completely specialized. Thus, for this school, the East
Asian miracle is driven by the rapid growth of factor endowments sustained by international trade,
3
and it can continue as long as the economies remain small and open.
In terms of empirical evidence, there is overwhelming support for the endowment-driven hy-
pothesis. Using aggregate (primal) growth accounting techniques to infer the growth of primal
total factor productivity (TFP), Young (1992, 1995) shows that most of the gross domestic prod-
uct (GDP) growth of the NIEs could be explained by their aggregate capital accumulation, such
that there is little sign of productivity growth in these economies. Young's results are supported
by many papers, including Kim and Lau (1994), Krugman (1994), Collins and Bosworth (1996),
and Kohli (1997).
It is only recently that the productivity-driven hypothesis has been resurrected by Hsieh (1999,
2002). Hsieh derives the implied (dual) productivity growth of the four East Asian NIEs based on
the market factor returns in these economies. He shows that the dual TFP growth is in general
higher than the primal TFP growth by 1 to 2 percentage points in these economies, depending on
the various measures of rate of returns to capital investment. The difference is especially large for
Singapore, which may have inflated aggregate capital investment data in its national accounts and
caused a smaller primal TFP growth. Hsieh attributes the discrepancy between the primal and
dual TFP growth rates to data issues. So far, this is the only piece of evidence that supports the
productivity-driven story.
The central idea of this paper is simple: If the contribution of productivity is indeed large at
the aggregate level, then we should find high productivity growth in the industries in the economy.
Conversely, if industry data show that most of the industry growth could be explained by the
growth of the aggregate endowments, then it would be consistent with an endowment-driven growth
hypothesis at the aggregate level.
To give some structure to the idea, we use a translog production-based GDP function approach
similar to Kohli (1991, 1997) and Harrigan (1997). We show that the contribution of an aggregate
endowment in GDP is correlated to the industry Rybczynski elasticity, which measures the per-
centage change of each industry output due to 1 percent increase in that aggregate endowment. On
4
the other hand, the aggregate contribution of productivity is correlated to industry productivity
growth. In both cases, the degree of correlation depends on the output share of the industry in
GDP.
We study the manufacturing and services industries of Hong Kong, which together cover more
than 99 percent of the economy, from 1984 to 1997.2 During that period, the GDP of Hong Kong
jtumped fivefold, and the aggregate capital stock more than doubled.3 At the same time, output
share of manufacturing in GDP dropped from 60 to 18 percent, share of import in GDP increased
by nearly 40 percent. The services sector, the remaining majority of the economy, was growing at
an average rate of 17 percent annually. A finding of a large and positive Rybczynski elasticity of the
services sector with respect to capital and a low productivity growth in the manufacturing sector
would be sufficient to reject the productivity-driven hypothesis in favor of the endowment-driven
hypothesis at the aggregate level.4
The results of our empirical analysis show that the services sector is indeed the more capital-
intensive sector, which benefited tremendously from the rapidly-growing capital endowment of
the economy. The estimated Rybczynski elasticity shows that for every 1 percent increase in the
capital endowment, output of the services sector increases by more than 2.4 percent. Given that
the average annual growth rate of capital endowment is nearly 8 percent, it fully explains all of the
output growth of the services sector in the sample period. Thus, even though the regression results
indicate that productivity elasticity of the services sector is positive and significant, given such a
large endowment effect, the role of productivity in the services sector is negligible. On the one
hand, productivity growth in the manufacturing sector is also found to be minimum - an average
' There are insignificant amounts of agriculture and fishing activities in Hong Kong.
3 The labor force of HonF Kong increased by only 20 percent during that period.
4 Given that the services share is growing when the aggregate capita endowment is increasing, a positive Rybczynski
elasticity of capital in the services sector would be consistent with the endowment-driven story at the aggregate level.
However, a positive Rybczynski elasticity of capital in the growing sector is not sufficient to lead to an aggregate
finding of endowment-driven growth, unless there is no sign of any industries' productivity growth in the economy.
To put it differently, if the Rybczynski elasticity of the services sector with respect to capital is not only positive,
it is large enough that most of the growth of the services sector is explained by the growing endowment and leaves
little sign of productivity growth, then to support the productivity-driven hypothesis at the aggregate level it would
be necessary for the manufacturing sector to have high productivity growth.
5
of 0.6 percent annually. The manufacturing sector is revealed to be more labor intensive. Its output
growth is predominantly hindered by the reallocation of production factors into the services sector
as a result of the growth of the aggregate capital endowments and imports.J
Combining a large and positive Rybczynski elasticity of the services sector with respect to the
growing capital endowment with a lackluster productivity growth in the manufacturing sector, this
paper concludes that there is sufficient sectoral evidence to reject the productivity-driven hypothesis
in favor of the endowment-driven hypothesis at the aggregate level in Hong Kong. The results are
robust to the possible endogeneity of industry productivity and the aggregate capital endowment.
This paper is organized as follows. A production-based GDP function that includes imports is
derived in Section 2. Empirical specification utilizing a translog funLction is developed in Section
3. The relationships between aggregate growth accounting and sectoral elasticities are presented in
Section 4. The data set used for the empirical analysis is shown in Section 5. The estimations and
results are discussed in Section 6. Robustness checks of the estimation are provided in Section 7.
The conclusion is presented in Section 8.
2. Theoretical Model: A General Equilibrium Setting
Consider a neoclassical small open economy with fixed aggregate factor supplies, constant returns
to scale production technology, and perfectly competitive goods and factor markets. This economy
has two main sectors, manufacturing and services; together there are N industries. Each indus-
try n produces only one good (Yn) from primary factors (vn) and intermediate materials (Zn).
Intermediate materials are sourced both domestically and from overseas. There are I kinds of pri-
mary factor in the economy. In each period t the production of each industry n is subjected to a
Hicks-neutral productivity progress, Ant. The GDP of the economy is equal to the total output of
the industry minus the value of imports, pmtMt. Given aggregate primary factor endowments (v),
the productivity level, and import and export prices, the general equilibrium of this small open
5 In other wvords, thc manufacturing sector is revealed to be more labor intensive, so it is hurt by the negative
Rybczynski effect as the economy becomes more capital abundant.
6
economy is obtained by reallocating resources to maximize its GDP, subject to all the production
and resources constraints:
N+1
max GDFt = E (p.t At) 9t
n=l
s.t. Yt = fn (Vnt, Znt) I n = 1, ...,N
YN+lt = -Mt
N
Evnt = vt, vtER', (1)
n=l
where for simplicity of presentation, we treat the negative of import demand as the (N + 1)th
output supply of the economy and let YN+1 = -AN+lMt, with AN+lt _ 1.
The assumption of constant returns to scale in production functions ensures that the second
order sufficient conditions for maximization hold. Hence the solution to the first order conditions
imply that GDP is a function of the prices of domestic output and imports, the sectoral productivity,
and the aggregate endowments:
GDFt = GDF (ptAt, vt) (2)
pt e R+
At= diag {Alt, A2t, ..., ANt, 1} E RN+1 x RN+1
Vt E R,
where * denotes the optimum. At is a diagonal matrix that defines the level of productivity of the
economy, and Pt is the price vector of the economy.° The second order sufficient conditions also
imply that GDF* is convex in pi, and At.7
The GDP function presented in Equation (2) incorporates two GDP function models developed in Kohli (1991,
1997) and Harrigan (1997). Kohli (1991) shows that the import price is important in explaining the expenditure-based
GDP function, and we can derive import demand from the GDP function. Harrigan (1997) introduces productivity
into the production-based GDP function by recognizing the multiplicative nature of prices and productivity in the
revenue function. This enables him to model productivity empirically, in a similar way as prices. Thus the current
GDP model includes the possible terms of trade effect or import competition faced by domestic industries, as well as
possible efficiency gain due to the relocation of the aggregate resources as a response to sectoral productivity shocks.
7 Notice that with the assumption of a small bpen economy, pt is exogenous and is-fixed in the world market. In the
context of a large economy, pt would depend on domestic output and would not enter the GDP function.
7
By the envelope theorem, the output supply of industry equals the gradient of GDF with
respect to own price, and import demand equals the negative of the gradient:
* t= GDF*(ptAt,vy) (ptAt,vt), Vn=l,...,N, (3)
Mt* = -yN+lt (ptAt,vt) = 8GDP* (ptAt,vt) (4)
Define the share of the output of industry n in GDP as snt = , then by construction, the
sum of all the industry's shares will be greater than 1, and the share of imports will be negative.
By Equation (3) it can be shown that the share of output of industry n in GDP is the elasticity of
GDP* with respect to its price:
Snt = 9 InGDP(ptAt, t) = sn(ptAt,vt), Vn= 1,...,N+1. (5)
N+1
s* t > 0, Vn =1..,N, sN+it < 0, E Sn = 1.
n=l
In addition, given the multiplicative nature of prices and productivity, for every industry n, the
elasticities of GDP with respect to pnt and Ant equalize:
a lnGDP* (ptAt,vt) _ IlnGDP* (ptAt,vt)
a lnpnt a In Ant
In other words, the share of industry n also equals the elasticity of GDP with respect to productivity
of n.
Hence in this general equilibrium framework, the share of industry n in GDP depends not only
on its own price and productivity, but also on the prices of all other goods, their productivity, and
the aggregate endowments of the economy.
With a similar method, we can also show that the share of factor i in total value added equals
the elasticity of total value added with respect to the quantity of i:
S* = anGDPt (6)
Our ultimate objective is to estimate the contributions of productivity and factor endowments
to output growth of the industries. One method would be to estimate the elasticities of output
8
with respect to productivity and factor endowments, and use the estimated elasticities to construct
the corresponding contributions.
Specifically, for every industry n and m, y*t = tDF and snt = InAnt . Given the
shares of n and m, the elasticity of n's output with respect to the productivity of m, eCA, is a
as,
linear function of the partial effect, -m
enmt8 A A = yn l At + s*} Vn,m ,..., + 1. (7)
Similarly, for every industry n and factor i, the factor elasticity of n with respect to i, ef is
also linear in the partial effect
Cf - 9l1nynt =1 9snvt +sit, Vn = 1,...,N+ 1, Vi = 1,...,. (8)
t5ln vit=*t a In vit+5t
The factor elasticity is known as the Rybczynski elasticity in the literature.
Finally, it can be shown that own price elasticity of each industry equals its own productiv-
ity elasticity minus 1, while cross price elasticity of each industry equals its corresponding cross
productivity elasticity:
Olnpnm4t,,f m
|En,nt, Vn ˘6 m
Thus, our empirical strategy would be first to estimate the partial effects of productivity and
factor endowments on the output shares, namely Cj and . Subsequently, we will construct
the elasticities using the corresponding estimated partial effects and shares, according to Equations
(7) and (8). Finally, for every industry n, output growth is decomposed as follows:
N+1 N+1 1
Yn*= E 4nAmt + E ePmt7Pnt + EfnitVit-
m=1 m=l i=l
3. Empirical Specification
To implement the model empirically, let us assume that GDP' (ptAt,vt) is a translog function
of productivity, prices, and factor endowments, with productivity and prices of goods entering
9
multiplicatively. Let n and m be the indices for industries and i and j be the indices for factors:
N+1 1N+1 N+1
lnGDF (ptAt,vt) = aoo + E ao7 In (Antp71t) + 2 E E anm In (A.tp.t) In (Amptmt)
n=1 .=I m=l
+E bo, In vit + - bij In vit In vjt (10)
i=1 i=1 j=1
N+1 I
+ E E cni In (Antpnt) In vit, (11)
n=1 i=1
with the usual symmetry and homogeneity restrictions:
anm =amn, bij = bji, Vn, m = l, ..., N + l, Vi, j I ,-.., I,
N+1 N+1 I
E aon 1, .a.n=, 5Cni=, N + 1,
n=1 mn=l =
I I N+1
boi= 1, Ebij = 0, E cni = 0, i =,...,I. (12)
i=1 j=1 n=1
Thus, the share of industry n in total value added can be derived as the elasticity of GDP*
with respect to pt based on Equation (10) and the above restrictions:
N+1 I
sn(ptAt,vt) =aon + Eanmn(Amtpmt)+ cniInvit, Vn= 1,...,N+1, (13)
m=l i=l
with anm and cni representing the partial effects of productivity and factor endowments on output
shares, j and n respectively. In other words, for every industry n, m, and factor i,
we can estimate the partial effects, anm and cni, by regressing output share of n on the levels of
productivity, price indices, and factor endowments, as shown in Equation (13).
However, two obvious problems are associated with the estimation of Equation (13). The first
problem is the non-stationarity of the level of productivity and prices, which causes the ordinary
least squares (OLS) estimates to be inefficient, as shown in Keller and Pedroni (1999). The second
problem is the lack of randomness of the model: With full information on the economy, Equation
(13) presents a complete model with no error term. Nevertheless, given that neither reliable data
on productivity nor prices of the services sector are easily available or constructible, empirically it
10
is impossible to have a full set of productivity and prices for all the industries in both the manu-
facturing and services sectors to implement Equation (13). By excluding prices and productivity
of the services sector from Equation (13), we introduce randomness to the model. On the other
hand, given that the partial effects, anm, and cni, are invariant over time, we can get around the
non-stationarity problem by taking the first difference of Equation (13).
Specifically, let the industry index for the services sector be n = 1. In order to capture the
highly non-stationary property of the level of productivity and price, we assume the log level of the
product of the productivity and the price of the service sectors follows a random walk with drift:
ln Aitpit = 6 + at + (t
(t = Ct-i + Ut, Ut - JV (O, a,O
Then by separating the services sector from the first summation of Equation (13), we have:
N+1 I
sn (ptAt, vt) = aon + anl (d + ^yt + (t) + E anm ln (Amtpmt) + rcni nvit, Vn= 1, ...,N+ 1,
m=2 i=l
(14)
with its first difference as
N+1 I
ds.* (ptAt, vt) = a,n (^t + Ut) + E anm (Ant + Pmt) + E cniit,
m=2 i=l
N+1 I
= an + E anm (Amt + Pmt) + cn t + Unt Vn = N +1, (15)
m=2 i=l
where an = anlY, unt = aniut, and the variable Zt denotes the growth rate of zx8 Equation (15)
shows that for every industry n, m, and factor i, the change in share of industry n, ds*, depends
on the growth rates of productivity, Amt, output prices, Pmt, and factor endowments, vit, as well
as an industry fixed effect, an.
Equation (15) can be further simplified by utilizing the dual definition of TFP,
Amnt -mt - Pmt, (16)
8 Specifically, tI nxt - Inzt1.
11
where tlimt denotes the weighted average of the growth rates of input prices.9 We can therefore
rewrite Equation (15) as
N+1 1
ds (ptAt,vt) =an+ E anmlwmt +EcniOit+unt, Vn= 1,...,N+ 1. (18)
m=1 i=1
Thus the change in share of industry n depends on the weighted averages of the growth rates of
input prices of all industries and the growth rates of factor endowments. Equation (18) will form
the basis of our estimation for anm and cni, Vn, m, i.
For every industry n, m, and factor i, the estimated productivity elasticity and the factor
elasticity are respectively
enAmt = + Smt, and (19)
Snt
enit= C*i+ t (20)
nt
4. Multisector Aggregate Growth Accounting
Equations (19) and (20) allow us to reinterpret the traditional aggregate growth accounting as the
output share weighted average of the sectoral productivity and Rybczynski elasticities:
N+1 N+1
Z S*tEAmt = Smt, E Snt,-f = S! (21)
n=1 n=1
In other words, the aggregate factor share equals the average Rybczynski elasticity of the economy,
and the industry share equals the average productivity (price) elasticity.
!) For cxample, if there are only four kinds of inputs, namely labor (L), capital (K), domestic materials (D), and
imported materials (M), with input prices equal to w, r, pD, and pM respectively, then
W.t= 0ntbn + 6ntt + n,tn + OntPt (17)
JL O 5 ajnLn + Wnt-IL.t-i
-t 0.5 * +
\ PntYnt Pnt-IYnt-I J
nt= 0.5* jn2Knj + rnt.lKnt-l)
= P +tnt pet-Ine.t-i
@° = O 5 :* (EL2tnt + pnot_ 1Dnt-l )
PntYnt pnt-IYnt-l
n't =0.5 * (EgEMLt + PntI- iMnt -I)
pntYUnt pnt-I Vnt-I
-L _ __ _ -D -All _
1 = snt + Ont + Ont + Ont
For the case of import, Wkt = Pkt since AAt is assumed to be 1.
12
It could be shown that under such a specification, the growth rate of GDP consists of the
following terms:
N+1 N+1 1
GDF (ptAt, vt) = Z s4tpnt + E s.tA.t + s stvit, (22)
n=1 n=1 i=l
where the first summation captures both the domestic price effect and the terms of trade effect a
la Kohli (1997). Utilizing the average elasticity interpretation of industry and factor shares, we
conclude that the growth rate of GDP depends on the growth rates of industry prices, industry
productivity, and aggregate factor endowments. The contribution of these determinants depends
on the average productivity and Rybczynski elasticities across all sectors.
Finally, according to this interpretation, the endowment-driven hypothesis is correct if the
majority sector of the economy has a large positive Rybczynski elasticity with respect to the
fast-growing endowment and has little productivity growth. Similarly, the productivity-driven
hypothesis is correct if the majority sector has large productivity growth and has small or negative
Rybczynski elasticity with respect to the fast-growing endowment.
5. Data
We aggregate the 26 industries in Hong Kong's manufacturing sector into five major manufacturing
industries. Together with imports and the services sector, there are a total of seven aggregate
industries. On the other hand, we only consider two types of primary aggregate factors, namely
labor and capital. Both labor and capital are homogeneous inputs.
Table 1 presents the concordance of the five major manufacturing industries to their Hong Kong
Standard Industrial Classifications. From 1976 to 1997, there were two classification regimes, with
the break taking place in 1990. Due to data reporting problems, the food, beverage, tobacco, and
petroleum and coal products industries (SIC 311/312, 313, 314, and 353/354) are excluded from
the sample. Data sources and the constructions of the variables are in the appendix.
Table 2 presents some summary statistics of the variables used in the regressions. It is clear
from the growth in real output and output share in GDP that the manufacturing sector as a whole
13
has been shrinking. Among the manufacturing industries, two of the largest industries in 1985,
textiles/machinery and electronics, each dropped from more than 20 percent of GDP in 1976 to
less than 7 percent in 1997. The rate of decline is rapid by any measure. On the other hand, the
aggregate factor endowments of the economy have been increasing. The growth rate of capital is
on average nearly 8 percent a year, while the growth rate of labor is 1.6 percent.
Evidence of import competition is clearly demonstrated by the growth rate of imports and
the change in the import share in GDP. While the value of manufacturing imported materials is
dropping due to the decline in manufacturing output, the imports as a whole is certainly getting
larger. In other words, we expect to see a lot of negative effects on the manufacturing industries
coming from imports.
Finally, the growth rate of productivity of the manufacturing sector shows sign of declining. This
is also true at an industry level for machinery and electronics and for miscellaneous manufactures.
6. Estimations and Results
Equation (23) shows a system of seven equations, and Equation (24) presents the 21 restrictions.
For each equation, the dependent variable is the change in share of output in GDP, with un being
the industry-specific error term. Notice that we are not imposing restrictions on the homogeneity
in prices in the system, as we do not have a complete set of prices.
7 2
dsnt = a+Zanmhmt+ZcniDitą+u nt, Vn=1,...,7 (23)
m=2 i=1
2
anm = am,, E Cni = O, Vn, m, i. (24)
m=1
Right-hand side variables for each equation include the weighted averages of the growth rates of
input prices of all the five industries plus import price, and the growth rates of the two aggregate
factor endowments. Given that the dependent variable is the change in share of output of one of the
seven industries in the sector for each equation, the error terms of the equations will be correlated
by construction. Hence the proper way to implement the empirical model will be to estimate it
14
as a system of six equations using iterative seemingly unrelated regressions.'0 Given that the
estimates are neutral to the dropping equation, without further complication, we choose to drop
the services sector out of our system and will recover its coefficients via symmetry and homogeneity
restrictions.
Table 3 presents the regression results of the system. Each of the six columns in the table
represents the regression result of one industry. The dependent variable of each regression is the
change in share of the industry in the column, and there are nine explanatory variables for each
regression. These explanatory variables are categorized into three groups. The first group consists
of the weighted averages of the growth rates of input prices of the various industries as well as the
growth rate of import prices. The second group of explanatory variables includes the growth rates
of the two aggregate factors. The third group is the industry fixed effects.
At first glance, most of the partial effects of productivity are estimated with precision, while all
of the partial effects of factor endowments are not significant. Moreover, all of the partial effects
of own price on output are positive and significant, and most of the partial effects of import price
on output of the industries are negative and significant. This finding is in line with the theoretical
restriction of the model.11
Table 4 shows the estimated productivity elasticities of the five manufacturing industries, the
services sector, and imports. Elasticities for the services sector are obtained by imposing the
symmetry and homogeneity restrictions. Each cell shows the percentage change in output of the
industry in the column due to a 1 percent change in productivity of the industry in the row.12
As shown in bold in Table 4, all of the estimated own productivity elasticities are positive and
significant. The range of the estimated own productivity elasticities is between 1.2 and 5.4. In
addition, all manufacturing industries have estimated own productivity elasticities that are signifi-
]This is equivalent to estimating a system using maximum likelihood estimators. See Barten (1969) for details.
i To satisfy convexity of GDP function, all the output supply of the industries should be positively related to own
price and negatively related to import price, where import is taken as the intermediate input. Thus it is necessary
that all the partial effects on own price be positive and partial effects on import price be negative.
IIFor example, a 1 percent increase in productivity in the textile industry causes the output or the chemicals industry
to decrease by 1.33 percent. It also leads to a 1.89 percent increase in the output of the machinery and electronics
industry.
15
cantly greater than 1. In other words, for each of the five industries in the manufacturing sector,
a 1 percent increase in own productivity will induce more than 1 percent increase in the output
of the industry. The productivity elasticity of the services sector is positive but not significantly
different from 1. Given that own price elasticity equals own productivity elasticity minus 1, the
regression result satisfies the specification of the theoretical model that the own price elasticities
should be nonnegative, as shown in Table 5.
Interestingly enough, imports react positively to productivity growth in the industry, even
though the estimated elasticities are less than unity. Thus, when there is technological progress
in the manufacturing sector, we would expect to see an increase in import demand. On the other
hand, as shown in Table 5 that all of the import price elasticities of the manufacturing industries
are negative and significant. For the manufacturing industries, a 1 percent increase in import prices
decreases industry output from 3.7 percent to 6.5 percent. For example, from 1984 to 1997, import
prices increased by more than 20 percent, and as a result, output dropped by 130 percent in the
miscellaneous manufactures industry. Thus the rising imports in the sample period have produced
some huge negative effects on the output of the manufacturing industries.
Table 6 presents the estimated factor elasticities. These elasticities are also known as the
Rybczynski elasticities, which measure growth of output due to the growth of the aggregate factor
endowments in an economy. Each cell shows the percentage change in output of the industry in
the column due to a 1 percent growth of the factor in the row.
According to Table 6, the estimated Rybczynski elasticity with respect to aggregate capital of
the services sector is positive and statistically significant. In other words, the services sector is
revealed to be capital intensive. For every 1 percent increase in the aggregate capital endowments,
output of the services sector increases by 2.4 percent. Given that, from 1984 to 1997, the average
annual growth rate of Hong Kong's aggregate capital endowments is nearly 8 percent, this would
cause the output of the services sector to increase by more than 18 percent annually. Thus, ac-
cumulation of capital endowments alone can explain all of the services sector's growth, leaving no
16
room for productivity growth in the sector.
On the other hand, most of the manufacturing industries are revealed to be labor intensive, with
positive Rybczynski elasticities with respect to aggregate labor endowment and negative Rybczynski
elasticities with respect to capital endowment. However, with the exception of the miscellaneous
manufactures industry, the elasticities are not precisely estimated. One possibility is that given
that most of the manufacturing industries have a concurrent decline in output over the sample
period, these elasticities are likely to be highly correlated, which make it difficult to estimate each
individual elasticity precisely. Nevertheless, given the strong positive Rybczynski elasticity with
respect to aggregate capital of the services sector, it is safe to infer that overall the manufacturing
sector is revealed to be labor intensive.
A labor intensive manufacturing sector would have benefitted from the increase in the aggregate
labor endowment. However, given that the average growth rate of the aggregate labor endowment
is only 1.6 percent annually, it helps little in offsetting the negative effect of the faster cumulating
aggregate capital on the manufacturing industries. Resources are moving into the services sector
from the manufacturing sector as a result of the changes in the mix of aggregate endowments which
push the economy to be more capital abundance.
Overall, the estimated productivity, prices and factor elasticities suggest that the growth of the
capital intensive services sector is mainly driven by the growth of the aggregate capital endowment,
while the growth of the labor intensive manufacturing sector is mainly hindered by the reallocation
of resources into the services sector as a result of the growth of the aggregate capital endowment
and the rising imports.
We also perform some specification tests on the regression results. All of the industries satisfy
the homogeneity hypothesis, which implies that the constant returns to scale assumption is not
rejected by the data. On the other hand, none of the industries satisfies the symmetry hypothesis.13
However, it is not unusual for such regularity conditions to fail in this type of model, and it
13Detailed results on the specification tests are available upon request.
17
is necessary to impose such restrictions for the estimation to conform to the model. Failure in
symmetry restriction could be due to the fact that the sizes of the industries are quite different,
ranging from 2 percent of GDP to 112 percent of GDP (including import). Harrigan (1997) has
similar findings in the system of equations of the OECD countries.
7. Robustness Checks
7.1 Endogeneity of TFP
There are at least two reasons why the sectoral growth rates of TFP and the contemporary regres-
sion errors could be correlated and cause the estimates to be biased. The first has to do with the
measurement of TFP, and the second reason is due to econometric issues associating with the fixity
of some inputs. Both of these issues will overestimate the industry productivity growth, leading to
underestimation of the productivity elasticities.
Specifically, the value of total industry output is used to construct the share of industry in GDP
and its changes. On the other hand, by invoking the dual definition of TFP, according to Equation
(16), we use data on total cost to construct the growth rate of industry TFP. With the assumption
of perfect competition, value of total output equals total cost. Hence we may have mechanically
introduced a spurious correlation between the dependent variable and the growth rates of industry
productivity.
In addition, if there is fixity of some inputs in the short run, then a sectoral-specific shock will
affect the sector's share in GDP and the measured sectoral TFP concurrently. This is similar to
the classical econometric problem of estimating a production function.
We use the lagged growth rate of the industry TFP as an instrumental variable to get around
the potential endogeneity issue of the current-period industry TFP growth. As a results, we use
the full information maximum likelihood estimation to fit the above system of equations. While the
point estimates of the regression are slightly different, they do not significantly alter the industry
productivity and Rybczynski elasticities. Correcting for endogeneity of productivity raises the
18
services sector productivity elasticity from 1.17 to 1.18. We maintain the earlier results that
growing capital endowment is the main driving force behind the growth of the services sector, while
the manufacturing industries are hurt by the reallocation of resources into the services sector due
to capital accumulation and import competition.14
7.2 Endogeneity of the Aggregate Capital Endowment
Hong Kong is one of the world's most open economies, not just in terms of movement of goods
and services but also in terms of movement of capital, both inward and outward. As such, the
aggregate capital stock of the economy could be a result of investors' response to the different rate
of returns across countries, as well as across industries. Specifically, a growing sector of a booming
economy provides investors with a higher expected rate of return in the future and further attracts
investment and causes the aggregate capital endowment to grow. This situation would lead to an
overestimation of the Rybczynski elasticity of the growing industry.
While the standard H-O model and Rybczynski theorem call for aggregate capital endowment to
be exogenous, with the free trade of goods and services, returns to factors are nevertheless equalized
across countries and sectors. Thus, we could use interest rates as an instrument of the aggregate
capital stock, which would capture the exogenous movement of capital due to changes in interest
rates that are not related to specific industries. A full information maximum likelihood estimation,
with the best lending rate of Hong Kong used as the instrument for the aggregate capital endowment
is performed. Once again, while the point estimates are slightly different from those in Table 3,
they do not change the industry productivity and Rybczynski elasticities significantly. Correcting
for the endogeneity of the aggregate capital endowment reduces the Rybczynski elasticity of the
services sector with respect to capital from 2.42 to 2.38. We maintain that the growth of the
services sector is predominantly due to the growth of the aggregate capital endowment.15
i 'Dctailcd regression results are available upon request.
niDctailcd regression results are available upon request.
19
8. Concluding Remarks
This paper sets out to find sectoral evidence that may substantiate the existing aggregate findings
in the literature regarding the relative importance of productivity and endowments in the growth
of Hong Kong.
Under a general equilibrium framework of a production-based GDP function approach, this pa-
per links the contributions of aggregate productivity and endowments to industry-level productivity
and Rybczynski elasticities. Given the drastic cumulation of aggregate capital stock, a finding of
a large Rybczynski elasticity of the majority sectors with respect to capital would be consistent
with the endowment-driven hypothesis. On the other hand, if most of the growth of the majority
sectors could be explained by factors other than productivity, then it would be inconsistent with
the aggregate productivity driven hypothesis.
The results of an iterative seemingly unrelated regression indicate that most of the growth of the
services sector is driven by the rapidly-accumulating capital endowments, and not by productivity
growth. In addition, productivity growth in the manufacturing sector is also unimpressive. The
manufacturing sector is revealed to be more labor intensive and it's growth is hindered by the
reallocation of its production factors into the services sector as a result of the growth of capital
endowments and imports. Overall, sectoral evidence supports the endowment-driven hypothesis.
The results are robust to the corrections of endogeneity of industry productivity and aggregate
capital endowment.
In terms of relevancy to the trade literature, this paper is the first to estimate the sectoral
Rybczynski elasticities and relate them to the aggregate growth of a small open economy. In terms
of relevancy to the growth literature, the sectoral evidence of this paper substantiates those existing
endowment-driven findings which so far have been mainly focused on the aggregate statistics.
20
A Appendix
Al Data Sources
Most of the industry-level raw data are from the Survey of Industrial Production published by
the Census and Statistics Department of Hong Kong from 1976 to 1997. Earlier year data are
supplemented by Hong Kong Annual Digest of Statistics, published by the same source. Data from
these sources include value of gross output (pntynt), value of materials purchased (pnDtDnt + PnMMnt),
number of persons engaged (Lnt), compensation of employees (wntLnt), gross addition to fixed assets
(value of investment: pInt).
Hong Kong Annual Digest of Statistics also provides data necessary for the construction of
the aggregate factor endowments, which include labor force (Lt) and gross domestic fixed capital
formation (value of aggregate investment: ptlIt).
Finally, the Census and Statistics Department of Hong Kong collected detailed Hong Kong trade
data at a commodity level from 1984 to 1998.11 This data set provides us with information on the
value and-quantity of import and export by commodities, year, and country of origin/consignment.
Given the highly disaggregate nature of the data, it is possible to construct unit value of import
and export by industry. Because trade statistics begin in 1984 and the industry data end in 1997,
this determined the time dimension of this paper.
A2 Capital Stock and Factor Shares
Both industry and aggregate real investments are inferred by deflating the value of investment
by the appropriate GDP deflator of gross domestic fixed capital formation. Capital input is then
compiled using the perpetual inventory method from real investment,
K = Knt* (-6) + It, (25)
loThc data are purchased by the Pacific Rim Business and Development program at the University of California at
Davis, and are only available for students and faculty of UC-Davis.
21
with the assumption that we correctly specify some base year level of capital stock, Kno. Fortu-
nately, the 1976 Survey of Industrial Production publishes the book value of all assets by industry.
Taking 1976 as our base year, we compile industry-level capital stock by Equation (25), at a fixed
depreciation rate of 10 percent. Log difference of industry capital stock gives us the growth rate of
industry capital input.
There are no published data for the aggregate capital stock in the base year. However, aggregate
investment series is available since 1972. We take 1972 as the base year to compile aggregate capital
stock. Given the high growth rate of aggregate investment, any underestimation at the beginning
of the series would not be significant for the later years, when we want to construct the growth
rate of aggregate capital stock, as we need only the growth rates of aggregate capital after 1984 for
regression purposes.
There are no published data on the shares of labor and capital in GDP of Hong Kong. Labor
share in GDP is constructed as a weighted average of industry's labor shares, with the share of
each industry in GDP as the weight, and capital share in GDP is constructed as 1 minus the labor
share:
ztLt =,N N VAnt wsntL.t
w~Lą - n=1 WntLnt _ VA26w)L~
GDPt GDPt LdGDPt VAnt (26)
Industries included in the construction of aggregate labor share are manufacturing, wholesale
and retail trades, restaurants and hotels, transport and related services, storage, communication,
financing and business services sectors, banking and insurance industries. All these industries
together account for more than 80 percent of the economy.
A3 Export and Import Prices
Export prices are constructed using Tornqvist price index from the unit value of export commodities:
dPtin I n Vn, t, (27)
i~= , nt' nt_ __ __ 1_
where in = 1,I...,nt is the group of common export commodities between year t and t - 1 in
industry n, and 9int is the average share of commodity in in the the total value of export of
22
industry n between year t and t - 1.17
9int = 0.5 * ( I , it + rEit , ), Vin, t. (28)
Sin= Pntnt in=l Ptlnt-1
Thus for every year and industry, we need to first identify the group of common export com-
moditics between last and current year, then construct the share of each commodity in the group
of common commodities for each of the two years, and take the average of the shares to obtain 9int'
Average share, 0int, is the weight used to construct the change in export price of industry n from
the change in log unit value of the commodities. In short, the change in industry price equals the
weighted average of the change in log unit value of the commodities in the industry. Import prices
are constructed in the same way.
There are three different commodity classifications used from 1984 to 1997. Commodities were
classified under 6 digits SITC revision 2 for 1984-1991, 6 digits SITC revision 3 for 1992-1993, and 8
digits HS for 1994-1998. We first tried to match up the commodities under different classifications
by the appropriate concordances. However, the generated price indices presented big swings in
1992 and 1994, which showed that the matching process was not successful. As such, in order to
minimize the noise in the data, changes in price of the industry for these two years were interpolated
from the rest of the years. Finally, with the help of a SITC to SIC concordance, all the commodities
are aggregated using the Equation (27) to construct industry-level export price indices.'8
A4 Domestic versus Imported Materials
To infer the values of domestic and imported materials from the value of total materials purchased,
we need to refer to the input-output tables of Hong Kong, which detail the composition of imported
and domestic materials by industry over time. Unfortunately, there is no frequent publication of
Hong Kong input-output tables other than those compiled by GTAP in 1995.19
I7Commoditics imported from different countries, or exporting to different countries, are considered as different
commodities.
isConcordances used in this paper can be found on the following web site maintained by Jon Haveman:
http://www.eiit.org/Trade.html.
i GTAP stands for the Global Trade Analysis Project, which was established in 1992 by Thomas Hertel at Purdue
University. It has a rich global database, which includes individual country input-output tables that account for
intersectoral linkages.
23
There are two ways we can make use of the information provided from the input-output table
of Hong Kong in 1995. The first is to assume that purchase shares of industry in total imported
materials stay constant. In other words, if in 1995, the textile industry purchased 35 percent of
the imported chemical products, then we assume that textile industry demands 35 percent of the
imported chemical products for all the years. Thus the change in the total import of chemical
products equals the changes of chemical products materials in all industries, regardless of the
intensity of the materials in production.
Alternatively, we assume that within each industry, the expenditure shares of various imported
materials in total imported materials stay constant. In other words, total imported materials of
each industry can be thought of as a Cobb Douglas function of the different types of imported
materials. Thus, if in 1995 the expenditure share of chemical materials in the total imported
materials of textiles was 13 percent, then we assume that the share of chemical materials in total
imported materials of textile industry stays at 13 percent for all years. In this way, an increase in
the imports of chemicals products will have a different impact on different industries, and the size
of the impact depends on the intensity of chemical materials of the industries. The same method
applies to domestic materials.20
We use the expenditure shares to construct the growth rates of total domestic materials, with
the assumption that growth rate of each type of domestic materials equals the growth rate of total
domestic sales of the industry in which the materials are originated.2' Growth rate of the share
of domestic materials in total materials is calculated as the difference between the growth rates of
total domestic materials (pDDt) and total materials ( Share of domestic materials in
total materials is constructed by compiling change in share of domestic materials, and the share of
imported materials in total materials is 1 minus the share of domestic materials: OD = !Dt *
-.D OD=D b
nt = PntDn- pntZnt, and 9 1 = exp (n)
20nDetailed data on the expenditure shares are available upon request.
2iDomestic sales of industry n is the difference between total output and exports.
24
REFERENCES
Barten, A. P. (1969). "Maximum Likelihood Estimation of a Complete System of Demand
Equations." European Economic Review, vol. 1, p. 7-73.
Collins, Susan, and Barry Bosworth (1996). "Economic Growth in East Asia: Accumulation
versus Assimilation." Brookings Papers on Economic Activity, no. 2, p. 135-191.
Findlay, Ronald (1996). "Modeling Global Interdependence: Centers, Peripheries, and Fron-
tiers." The American Economic Review, vol. 86, no. 2, p. 47-51.
Harrigan, James (1997). "Technology, Factor Supplies, and International Specialization: Esti-
mating the Neoclassical Model." The American Economic Review, vol. 87, no. 4, p. 475-494.
Hsieh, Chang-Tai (1999). "Productivity Growth and Factor Prices in East Asia." The Amer-
ican Economic Review, vol. 89, no. 2, p. 133-138.
Hsieh, Chang-Tai (2002). "What Explains the Industrial Revolution in East Asia? Evidence
From the Factor Markets." The American Economic Review, vol. 92, no. 3, p. 502-526.
Keller, Wolfgang, and Peter Pedroni (1999). "Does Trade Affect Growth? Estirnating R&D-
Driven Models of Trade and Growth at the Industry Level." Draft paper prepared for the
conference on The Role of Technology in East Asian Economic Growth, UC-Davis, August
1999.
Kim, Jong-Il, and Lawrence Lau (1994). "The Sources of Economic Growth of the East Asian
Newly Industrialized Countries." Journal of the Japanese and International Economies, vol.
8, no. 6, p. 235-271.
Kohli, Ulrich (1991). Technology, Duality, and Foreign Trade: The GNP Function Approach
to Modeling Imports and Exports, Harvester Wheatsheaf.
Kohli, Ulrich (1997). "Accounting for Recent Economic Growth in Southeast Asia." Review
of Development Economics, vol. 1, no. 3, p. 245-256.
Krugman, Paul (1994). "The Myth of Asia's Miracle." Foreign Affairs, vol. 73, no. 6, p. 62-78.
Lucas, Robert E. Jr. (1993). "Making a Miracle." Econometrica, vol. 61, no. 2, p. 251-272.
Lucas, Robert E. Jr. (1988). "On the Mechanics of Economic Development." Journal of Mon-
etary Economics,- vol. 22, p. 3-42.
Ventura, Jaume (1997). "Growth and Interdependence." The Quarterly Journal of Economics,
vol. 112, no. 1, p. 57-84.
Young, Alwyn (1991). "Learning By Doing and the Dynamic Effects of International Trade."
The Quarterly Journal of Economics, vol. 106, p. 369-406:
Young, Alwyn (1992). "A Tale of Two Cities: Factor Accumulation and Technical Change in
Hong Kong and Singapore." NBER Macroeconomics Annual 1992. p. 13-53.
Young, Alwyn (1995). "The Tyranny Numbers: Confronting the Statistical Realities of the
East Asian Growth Experience." The Quarterly Journal of Economics, vol. 110, no. 3, p.
641-668.
25
Table 1: Data Description
Years: 1981-1998
Product classification system: There are a total of 5 industries, briefly follows the nine categories of the two-digit
level of the Hong Kong Standard Industrial Classification (HSIC). The categories, and their three-digit
HSIC constituent parts, are listed below.
Industry HSIC(81-89) Description HSIC(90-96) Description
Textiles 320/322 Wearing Apparel 320/322 Wearing Apparel
323 Leather Products 323 Leather Products
324 Footwear 324 Footwear
325-329 Textiles . 325-329 Textiles
Paper & 341 Paper Products 341 Paper Products
Printing 342 Printing & Publishing 342 Printing & Publishing
Chemicals 351/352 Chemical Products 351/352 Chemical Products
355 Rubber Products 355 Rubber Products
356 Plastic Products 356 Plastic Products
361-369 Non-Metallic Mineral 361-369 Non-Metallic Mineral
Machinery & 371/372 Basic Metal 371/372 Basic Metal
Electronics 380/381 Fabricated Metal 380/381 Fabricated Metal
382 Machinery 382 Office Machinery
383 Electrical, Electronic Products 383 Radio, TV &
Communication Equipment
384 Electrical, Electronic Parts 384 Electronic Parts
385 Scientific Instruments 385 Electrical Appliances
386/387 Machinery
388 Transport Equipment
389 Transport Equipment 389 Scientific Instruments
Miscellaneous 331 Wood Products 331 Wood Products
Manufacture 332 Fumiture 332 Fumiture
390/391 Other Manufacturing 390/391 Other Manufacturing
Share of each industry in total output of manufacturing sector
Source: Survey of the Census of Industrial Production, Hong Kong (SIP)
Prices of goods
Measured by Tornqvist unit value of exports, 1984-1998
Source: Census and Statistical Department, Hong Kong Special Administrative Region
Growth rate ofproductivity
Measured by the growth rate of dual TFP, which equals to the weighted average of the growth rates
of input prices minus the growth rate of output price. Source: SIP
Capital
Generated by compiling real investment using the perpetual inventory method with a depreciation rate of 10%
Source: SIP
Labor
Number of Workers
Source: SIP
notc: * HSIC 353/354 (Pctroleum and Coal Products) is not included due to the lack of data for the first half
of our sample.
26
Table 2: Data at a Glance
Paper & Machinery & Miscellaneous
Variables Years Manufacturing* Textiles Printing Chemicals*' Electronics Manuf6ctures
Growth rate of 1985 -6.0682 4.2574 -3.6711 -9.9102 -7.0832 -5.9565
real output 1997 -6.6898 -7.0543 10.2951 0.9745 -10.2835 -15.2646
mean -1.6259 -2.8623 8.2323 -1.5203 -3.4195 -5.9101
Output share in 1985 63.3975 26.0510 3.6012 8.0945 23.1072 2.5435
GDP 1997 18.3703 5.7102 3.0152 1.8082 6.9176 0.9191
mean 46.6110 17.7564 3.5614 5.1490 18.0762 2.0681
Change in 1985 -9.4480 -3.0853 -0.3068 -1.1359 4.3124 -0.6076
output share 1997 -3.3320 -0.9576 -0.0969 -0.2837 -1.8875 -0.1064
mean -4.1904 -1.8020 -0.0687 -0.5709 -1.5771 -0.1717
Growth rate of 1985 -5.1498 -5.0099 -0.9816 -4.3604 -7.0656 0.7698
labor input 1997 -12.9138 -20.0259 -3.3155 -2.4379 -13.0069 -10.6848
mean -9.6201 -11.8160 0.1375 -12.4491 -9.5455 -6.7535
Growth rate of 1985 1.0403 -1.4933 5.7958 4.2795 1.5403 2.4435
capital input 1997 0.0670 -7.1892 5.1020 1.0617 3.2378 4.0387
mean 1.9914 -2.1438 8.9152 0.6080 3.7834 2.6861
Growth rate of 1985 -0.0183 -0.1383 1.2950 -0.0416 -2.6897 1.2950
domestic materials 1997 -0.6184 1.0808 -2.4948 -3.5613 -3.5156 -2.4948
mean -4.1495 -10.4167 5.3958 1.3159 0.7093 5.3958
Growth rate of 1985 -2.4893 4.1282 -1.7080 4.1119 -1.0342 -1.7080
imported materials 1997 -3.3262 0.0494 -5.1927 -3.8111 -5.8325 -5.1927
mean -0.3556 0.7460 5.6183 -2.9746 -2.4599 -2.1475
Growth rate of 1985 -1.2056 -0.3177 2.1132 3.3962 -3.4111 -8.8478
output price 1997 0.1052 1.6337 -3.3732 -5.4577 -3.7528 14.3940
mean 3.8458 3.1427 2.5895 1.8035 5.6426 9.2487
Growth rate of 1985 0.6145 -0.9625 0.4368 -5.3264 2.8227 8.4632
productivity** 1997 -0.8846 0.4558 6.5895 4.6663 -1.7500 -17.1295
mean 0.5967 1.3408 2.4033 2.1025 -1.1075 -5.9419
Aggregate
Endowments Labor Capital Imports
Growth 1985 0.7911 7.1454 3.5405
rates 1997 3.8738 9.4722 5.0481
mean 1.6173 7.8511 15.2178
Share in 1985 45.8083 54.1917 86.8899
GDP 1997 42.8246 57.1754 122.4305
mean 42.2041 57.7959 111.9986
Notes: All values are in percentage terms. Mean values are the annual averages for the period 1984-1997.
*SIC 311/312,313,314 (Food, Beverage and Tobacco Products) are excluded.
'* SIC 353/354 (Petroleum and Coal Products) is excluded from the data due to the lack of data prior to 1988.
***productivity is measured as the dual total factor productivity.
27
Table 3: Dependent Variables: Change in share of output in GDP
Estimation method: Restricted Iterative Seemingly Unrelated Regression (MLE)
Total number of restrictions: 21
Total system observations: 78
Eq(l) Eq(2) Eq(3) Eq(4) Eq(5) Eq(6)
Independent Paper & Machinery & Miscellaneous
Variables: Textiles Printing Chemicals Electronics Manufacture Imports
Textile 03612*** 0.0232 -0.0776' 0.3093'* 0.0464*** 0.6967**
(0.0975) (0.0211) (0.0327) (0.0969) (0.018) (0.3102)
, Paper & 0.0232 0.1133*** -0.06470** 0.1 145*** 0.0370** 0.0927
Printing (0.0211) (0.0163) (0.022) (0.0251) (0.013) (0.0588)
a
< Chemicals -0.0776* -0.0647**' 0.2741*** 0.0183 0.0157 0.2517**
(0.0327) (0.022) (0.0329) (0.0383) (0.0176) (0.102)
2 Machinery& 0.3093*' 0.11450*0 0.0183 0.5196*** 0.0062 0.8368**
c Electronics (0.0969) (0.0251) (0.0383) (0.1283) (0.0238) (0.3759)
0
u Miscellaneous 0.04640** -0.037** 0.0157 0.0062 0.0705*** 0.1123**
, Manufactures (0.018) (0.013) (0.0176) (0.0238) (0.0169) (0.0486)
1Imports -0.6967** -0.0927 -0.2517* -0.83680* -0.1 1230* -1.1299
(0.3102) (0.0588) (0.102) (0.3759) (0.0486) (1.465)
* Capital -0.264 -0.0344 0.076 -0.3376 -0.0815 2.4095
U (0.3812) (0.0743) (0.1229) (0.426) (0.052) (1.794)
Labor 0.264 0.0344 -0.076 0.3376 0.0815 -2.4095
(0.3812) (0.0743) (0.1229) (0.426) (0.052) (1.794)
Industry Fixed -0.0198 -0.0051 -0.0116 -0.0248 0.0018 -0.1935*
Effect (0.0223) (0.0042) (0.0071) (0.0251) (0.0029) (0.1096)
Note: All figures in bold are the own partial effects of productivity. Standard errors are in parentheses.
*, ", and *** indicate significance at 90%, 95%, and 99% confidence levels respectively.
28
Table 4: The Elasticity of Output with respect to Productivity
Effect In terms of percentage change in output in:
Paper & Machinery & Miscellaneous
Tcxtiles Printing Chemicals Electronics Manufactures Service Imports
Textiles 2.2116*** 0.8293 -1.3294" 1.8884*'* 2.4221"** 0.1982* 0.7996*"'
(0.549) (0.5927) (0.6349) (0.536) (0.8689) (0.0975) (0.277)
Paper& 0.1663 3.2178** -*1.221l*" 0.6688*** -1.7544*'* 0.0014 0.1184**
S Printing (0.1189) (0.4568) (0.4263) (0.139) (0.6305) (0.0194) (0.0525)
. Chemicals -0.3855* -1.7653"' 5.3741**" 0.1529 0.8093 0.1034*** 0.2762"'
(0.1841) (0.6163) (0.639) (0.2119) (0.852) (0.033) (0.0911)
a.
.E Machinery& 1.9224"'e 3.3944"' 0.5367 3.055*** 0.4828 0.1015 0.9279"'*
Electronics (0.5456) (0.7053) (0.744) (0.7098) (1.1484) (0.1143) (0.3356)
, Miscellaneous 0.2821"' -1.0188"' 0.3251 0.0552 3.4294--- 0.027' 0.1209*"
; Manufactures (0.1012) (0.3661) (0.3422) (0.1314) (0.8152) (0.0151) (0.0433)
Service 1.8465" 0.0664 3.322"* 0.9289 2.1599* 1.1682*"* 0.8858
(0.9084) (0.9019) (1.0605) (1.0454) (1.2094) (0.4357) (0.7957)
Note: Figures in bold are the own productivity elasticities. Standard errors are in parentheses.
The productivity elasticity of industry n with respect to industry k equals the share of industry k plus
the ratio of the corresponding estimated cross partial effect to the share of industry n.
and *** indicate significance at 90%, 95%, and 99% confidence levels respectively.
Table 5: The Elasticity of Output with respect to Prices
Effect In terms of percentage change in output In:
Paper & . Machinery & Miscellaneous
Textiles Printing Chemicals Electronics Manufactures Service Imports
Textiles 1.2116** 0.8293 -1.3294** 1.8884*** 2.4221"0' 0.1982" 0.7996"'
(0.549) (0.5927) (0.6349) (0.536) (0.8689) (0.0975) (0.277)
Paper& 0.1663 2.2178"*' -1.221"' 0.6688*** -1.7544**" 0.0014 0.1184*0
Printing (0.1189) (0.4568) (0.4263) (0.139) (0.6305) (0.0194) (0.0525)
- Chemicals -0.3855" -1.7653*"' 4.3741*** 0.1529 0.8093 0.1034"** 0.2762**'
.6, (0.1841) (0.6163) (0.639) (0.2119) (0.852) (0.033) (0.0911)
.E Machinery& 1.9224*** 3.3944*** 0.5367 2.055"** 0.4828 0.1015 0.9279***
| Electronics (0.5456) (0.7053) (0.744) (0.7098) (1.1484) (0.1143) (0.3356)
A Miscellaneous 0.2821"' -1.0188"** 0.3251 0.0552 2.4294*"* 0.027* 0.1209***
a Manufactures (0.1012) (0.3661) (0.3422) (0.1314) (0.8152) (0.0151) (0.0433)
Service 1.8465"* 0.0664 3.322*** 0.9289 2.1599' 0.1682 0.8858
(0.9084) (0.9019) (1.0605) (1.0454) (1.2094) (0.4357) (0.7957)
Imports -5.0435"** -3.7239** -6.0075"** -5.7492"'* -6.5491*"* -0.5999 -3.1288"*
(1.747) (1.6522) (1.981) (2.0795) (2.3476) (0.5389) (1.3081)
Note: Bold face figures are own price elasticities. Standard errors are in parentheses.
All the cross price elasticities equal to the corresponding cross productivity elasticities,
while the own price elasticities equals to own productivity elasticities minus one.
',",and "' indicate significance at 90%, 95%, and 99% confidence levels respectively.
29
Table 6: The Elasticity of Output with respect to Factors
Effect in terms of percentage change in output in:
Paper & Machinery & Miscellaneous
Tcxtiles Printing Chemicals Electronics Manufactures Service Imports
Z Labor 1.9087 1.3877 -1.0537 2.2895 4.3631* -1.4227 -1.7293
Endowment (2.1468) (2.0863) (2.3869) (2.3567) (2.5158) (0.9128) (1.6018)
- Capital -0.9087 -0.3877 2.0537 -1.2895 -3.3631 2.4227*** 2.7293*
S Endowment (2.1468) (2.0863) (2.3869) (2.3567) (2.5158) (0.9128) (1.6018)
Note: Standard errors are in parentheses. The factor elasticity of industry n with respect to factor m equals the share of
factor m plus the ratio of the corresponding estimated partial effect and the share of industry n.
* *, and *** indicate significance at 90%, 95%, and 99% confidence levels respectively.
30
Policy Research Working Paper Series
Contact
Title Author Date for paper
WPS2868 Universal(ly Bad) Service: George R. G. Clarke July 2002 P. Sintim-Aboagye
Providing Infrastructure Services Scott J. Wallsten 38526
to Rural and Poor Urban Consumers
WPS2869 Stabilizing Intergovernmental Christian Y. Gonzalez July 2002 B. Mekuria
Transfers in Latin America: David Rosenblatt 82756
A Complement to National/ Steven B. Webb
Subnational Fiscal Rules?
WPS2870 Electronic Security: Risk Mitigation Thomas Glaessner July 2002 E. Mekhova
In Financial Transactions-Public Tom Kellermann 85984
Policy Issues Valerie McNevin
WPS2871 Pricing of Deposit Insurance Luc Laeven July 2002 R. Vo
33722
WPS2872 Regional Cooperation, and the Role Maurice Schiff July 2002 P. Flewitt
of International Organizations and L. Alan Winters 32724
Regional Integration
WPS2873 A Little Engine that Could ... Liesbet Steer August 2002 H. Sutrisna
Domestic Private Companies and Markus Taussig 88032
Vietnam's Pressing Need for Wage
Employment
WPS2874 The Risks and Macroeconomic David A. Robalino August 2002 C. Fall
Impact of HIV/AIDS in the Middle Carol Jenkins 30632
East and North Africa: Why Karim El Maroufi
Waiting to Intervene Can Be Costly
WPS2875 Does Liberte=Egalite? A Survey Mark Gradstein August 2002 P. Sader
of the Empirical Links between Branko Milanovic 33902
Democracy and Inequality with
Some Evidence on the Transition
Economies
WPS2876 Can We Discern the Effect of Branko Milanovic Au(gust 2002 P. Sader
Globalization on Income Distribution? 33902
Evidence from Household Budget
Surveys
WPS2877 Patterns of Industrial Development Raymond Fisman August 2002 K. Labrie
Revisited: The Role of Finance Inessa Love 31001
WPS2878 On the Governance of Public Gregorio Impavido August 2002 P. Braxton
Pension Fund Management 32720
WPS2879 Externalities in Rural Development: Martin Ravallion August 2002 C. Cunanan
Evidence for China 32301
WPS2880 The Hidden Costs of Ethnic Conflict: Soumya Alva August 2002 T. Bebli
Decomposing Trends in Educational Edmundo Murrugarra 39690
Outcomes of Young Kosovars Pierella Paci
WPS2881 Returns to Investment in Education: George Psacharopoulos September 2002 N. Vergara
A Further Update Harry Anthony Patrinos 30432
WPS2882 Politically Optimal Tariffs: Dorsati Madani September 2002 P. Flewitt
An Application to Egypt Marcelo Olarreaga 32724
Policy Research Working Paper Series
Contact
Title Author Date for paper
WPS2883 Assessing the Distributional Impact B. Essama-Nssah September 2002 0. Kootzemew
of Public Policy 35075
WPS2884 Privatization and Labor Force Alberto Chong September 2002 H. Sladovich
Restructuring around the World Florencio Lopez-de-Silanes 37698
WPS2885 Poverty, AIDS, and Children's Martha Ainsworth September 2002 H. Sladovich
Schooling: A Targeting Dilemma Deon Filmer 37698
WPS2886 Examining the Feasibility of Jerry R. Skees September 2002 E. Laguidao
Livestock Insurance in Mongolia Ayurzana Enkh-Amgalan 82450
WPS2887 The Demand for Commodity Alexander Sarris September 2002 M. Fernandez
Insurance by Developing Country 33766
Agricultural Producers: Theory and
an Application to Cocoa in Ghana
WPS2888 A Poverty Analysis Macroeconomic Luiz A. Pereira da Silva September 2002 R. Yazigi
Simulator (PAMS) Linking Household B. Essama-Nssah 37176
Surveys with Macro-Models Issouf Samake
WPS2889 Environmental Performance Rating Hua Wang September 2002 Y. D'Souza
and Disclosure: China's Green- Jun Bi 31449
Watch Program David Wheeler
Jinnan Wang
Dong Cao
Genfa Lu
Yuan Wang
WPS2890 Sector Organization, Governance, Antonio Estache September 2002 G. Chenet-Smith
and the Inefficiency of African Water Eugene Kouassi 36370
Utilities
WPS2890 Sector Organization, Governance, Antonio Estache September 2002 G. Chenet-Smith
and the Inefficiency of African Water Eugene Kouassi 36370
Utilities
WPS2891 Trends in the Education Sector from Nga Nguyet Nguyen September 2002 E. Khine
1993-98 37471