1

A CROSS SECTION ESTIMATE OF TRANSLOG PRODUCTION

FUNCTION: JORDANIAN MANUFACTURING INDUSTRY.
Ali M. Khalil
Al-Ahliyya Amman University
ABSTRACT
The objective of this paper is to estimate the Transcendental Logarithmic
Production
Function of manufacturing industry in Jordanian economy. Manufacturing
industry
can be considered as the fourth large one in Jordanian economy. In last
decade, the
growth rate in this sector is the second one after transport and
communications. The
gross fixed capital formation in manufacturing is the third one. In this
paper we need
a cross sectional data for cost shares, factor inputs, factor prices, and
output. Main
source of data for manufacturing industry in Jordan is the Industry Survey
-
Department Statistics - the Hashemite Kingdom of Jordan. The data for
factor inputs
and output were transferred to logarithm. A point around which
production function
should be estimated was chosen. Cost shares of inputs have been
calculated by
dividing compensations to employees, operating surplus, and value of
materials by
value manufacturing output. The theoretical framework in this research is
the flexible
production function. S0, the thee-input transcendental logarithmic
(Translog)
production function could be used. This function is approximated by
second order
Taylor series. The log of likelihood ratio test has been used to choose
among different
hypotheses. The symmetric translog hypothesis has been found as an
appropriate one
among other hypotheses. It is well behaved, where positivity and
concavity of
production function have been satisfied. After testing for nested
hypotheses, the
symmetry restrictions were be imposed. So, restricted and unrestricted
functional
forms are estimated. However, Iterative Zellner-Efficient Estimate has
been used to
get estimates equivalent to the maximum likelihood estimates. Allen
partial
elasticities of substitution, AES, and own and cross price elasticities for
factor inputs
have been calculated. AES measure the substitutability or complement
among factors
of production. On the other hand, AES can measure the curvature of
isoquant. AES
shows that capital-labor, capital–materials, and labor-materials are all
substitutive.
But, AES for labor-materials is more than a half of that for capital-labor.
AES for
capital-labor is one and three fourth of that for labor-materials. The Price
elasticities
of factor inputs show that capital and labor demand are more elastic than
demand for
materials.
JEL Classification Codes: D2
Keywords: Translog Production, Nonhomothetic Translog cost, AES, Factor
Price
Elasticity
2
_. INTRODUCTION
Jordanian manufacturing industry can be considered as fourth largest
sector in
Jordanian economy. In this paper we introduce the results of an attempt to
characterize the structure of technology in Jordanian manufacturing in
2002. This
paper can be considered as a first attempt to estimate the three-input
translog
production function for four digits seventy seven Jordanian manufacturing
industries
in 2002. In addition, we estimate partial elasticity of substitution, and own
and cross
price elasticities of factor inputs. The homothetic and nonhomothetic cost
functions
had been estimated. Then, our estimates of partial elasticities of
substitution and
factor cross price elasticities are based on IZEF estimates of homothetic
symmetric
translog cost function and the fitted cost shares of factor inputs.
Technology can be represented by a production function, such as Cobb-
Douglas and Constant Elasticity of Substitution (CES) place a restriction on
elasticity
of substitution (Cobb and Douglas 1928), (Arrow, et al. 1961).
Alternatively, the
generalized Leontief, generalized Cobb-Douglas and Translog function all
are
sufficiently flexible. The Translog function allows for variability of Allen
partial
elasticities of substitution and for using any number of inputs (Berndt and
Christensen
1973), (Caves and Christensen 1980). Bernd and Wood, in their estimate
of
homothetic constant returns to scale cost function, show that Allen
elasticities of
substitution (AES) are stable over time, but they are different from one
(Berndt and
Wood 1975). Christensen and Greene, estimate a nonhomothetic cost
function
(Christensen and Greene 1976). Bbani Hani and Shamia, had estimated
two-input
Cobb-Douglas production function for Jordanian industrial sector 1967-86
the found
that production function were decreasing returns to scale over the period
of study
(Bani Hani and Shammia 1989).
_. PRODUCTION FUNCTION:
The translog function is an attractive flexible function. This function has
both linear
and quadratic terms with the ability of using more than two factor inputs.
This
function can be approximated by second order Taylor series (Christensen,
et al. 1973).
The three-input translog production function can be written in terms of
logarithms as
follows,
Ln Q = _o +_K Ln K+ _L Ln L+ _M Ln M+
½ _KK Ln K2 + _KL Ln K Ln L + _KM Ln K Ln M
½ _LL Ln L2 + _LM Ln L Ln M + ½ _MM Ln M2 (1)
Where Q is the gross manufacturing output, K is real stock of capital
input, L is labor
input, and M is material input. _o is the intercept or the constant term. _K ,
_L , and
_M are first derivatives. _KK , _LL, and _M M are own second derivatives. _KL ,
_KM, and _LM are cross second derivatives.
3
Under perfect competition assumption, output elasticity with respect to
input
equals to cost share of that input. Thus, we can get a system of equations
from
differentiating the translog production function with respect to each factor
input,
_ Ln Q/ _ Ln K = _K + BKK Ln K+ BKL LnL+ BKM Ln M
_ Ln Q/ _ Ln L = _L + BLK Ln K+ BLL LnL+ BLM Ln M
_ Ln Q/ _ Ln M = _M + BMK Ln K+ BML LnL+ BMM Ln M (2)
Where BK represents the average cost share of capital, _KK , _KL, and _KM
represent
constant capital share elasticity with respect to capital, capital share
elasticity with
respect to labor, and capital share elasticity with respect to material input
respectively.
_MK _ML and _MM are constant material share elasticity with respect to
capital, with
respect to labor and with respect to materials.
Translog production function can be estimated under different nested
hypotheses. So, to choose among nested models, we use log of likelihood
ratio that
approximated by Chi-square with a degrees of freedom equal to number
of parameter
restrictions (Norsworthy and Malmquist 1983). The hypotheses to be
tested are: H0 :
P1’s = P2’s and H1 : P1’s _ P2’s. Where, H0 and H1 represent the null and the
alternative
hypothesis respectively. P1’s and P2’s represent parameter estimates from
unrestricted
and restricted model respectively. The test statistic is based on the
likelihood ratio.
This ratio is the maximum value of the likelihood function for restricted
production
function to the maximum value of the likelihood function for the
unrestricted one.
However, this test statistic is based on minus twice the logarithm of the
likelihood
ratio, -2Ln (R-U). Where, Ln R and Ln U represent restricted and
unrestricted log of
likelihood production function respectively. [1] However, under the null
hypothesis,
this test statistic is distributed asymptotically as Chi-square distribution
with degrees
of freedom equal the number of restrictions that we are testing. To choose
among
nested model we calculate Chi-square. Then, if we get a calculated Chi-
square less
than tabulated one, we accept the null hypothesis and we conclude that
restricted
model is appropriate to our data. But, if test statistic is greater than
critical value of
Chi-Square, we reject the null hypothesis and we say that the restricted
model is
inappropriate to our data.
However, we check for symmetry, constant returns to scale, existence of
separability, and even for Cobb-Douglas hypothesis (Berndt and Wood
1975),
(Norsworthy and Malmquist 1983).
So we impose symmetry restriction on parameters,
BKL =BLK ,
BKM = BMK
BML = BLM .
For constant returns to scale, we impose the following restriction:
4
_K + _L + _M =1,
BKK + BLK + BMK = 0,
BKL + BLL + BML = 0,
BKM + BLM + BMM = 0 .
And for weak separability, we check whether the linear separability
restrictions are
satisfied with Jordanian manufacturing data. The linear restriction might
be:
BLM = BMK = 0
Finally, we impose restrictions for existence of Cobb-Douglas,
BKK = BKL = BKM = BLL = BLM = BMM = 0 .
However, if log of maximum likelihood ratio is greater than critical value
of
Chi-square weak separability is rejected, and we conclude that value
added
specification is inappropriate to our data. After choosing the appropriate
model for our
data, we check for positivity of the function. Positivity is satisfied when
the fitted cost
share of capital, of labor and of materials are positive. Then, concavity of
the function
should be checked. Concavity of the function is satisfied if the Hessian
determinant of
IZEF parameter estimates is negative semidefinite (Berndt and Wood
1975).
Finally, we can measure the curvature of isoquant by estimating Allen
partial
elasticities of substitution (AES). However, it is possible to get AES for pair
of factor
inputs. It will be estimated from translog production function by the
formula,
_ fi xi | Bij |
_ij = ------------ . --------- ,
xi xj | B |
Where i and j represent inputs and they are different, i _ j . fi, | B | is and | Bji |
are the first
derivative, the determinant of bordered Hessian of the estimates, and the
cofactor of
the Bji parameter respectively. By Shephard duality, we can get ASE from
estimated
parameters of cost function and fitted estimated cost share of inputs as
follows,
_ij = ( _ij + Si S j ) / ( Si S j )
_ii = ( _ii + Si 2 – Si ) / ( Si
2) . (3)
Where _ii represent the estimated second order derivatives on the
diagonal of Hessian
Matrix. _ij represent parameter estimated of constant elasticities of cost
share with
respect to price of factor input service. Si and S j represent fitted cost share
of inputs
(Christensen, et al. 1971), (Diewert 1971), (Uzawa 1962). So, we can
recognize the
existence of complementary or substitutability among factor inputs of
production.
Factor price elasticities are related to AES as we see below:
_ij = ( _ij + Si S j ) / Si
5
_ii = ( _ii + Si 2 – Si ) / Si where i and j are K, L, and M. (4)
We should remember that, in general, _ij _ _ ji . [2]
_. ESIMATION OF PRODUCTION FUNCTION
The main sources of cross section data for Jordanian manufacturing is the
Department
Of Statistics (DOS), and the Central Bank Of Jordan ( CBJ ) (Central Bank of
Jordan) (Department of Statistics 2002). However, we have got data on
gross output,
gross value added by economic activity, data on number of workers,
capital stock,
depreciation of capital, and total fixed capital formation. In addition, we
get data on
compensation of employees by economic activity, operating surplus and
intermediate
goods and services that are used in production. Data on price of factors
such as price
of labor, price of capital and price of material input are obtained from
DOS.
To get the values of estimates for three-input manufacturing symmetric
translog production, we have to estimate the system of share equations
(2). This can
be done by OLS, whereas, a greater efficiency might be obtained by using
Zellner
Efficient (ZEF) estimation. So, we drop one of the three equations and we
estimate
only two equations. But, the problem with ZEF raise when we arbitrarily
drop one of
the equation in the system (2), where the estimates may not invariant
with respect to
the deleted equation. Thus, to avoid this problem we iterate ZEF until the
estimates
converge to the maximum likelihood ML estimates (Kmenta and Gilbert
1968).
However, we use IZEF to estimate the symmetric translog production
function. Then, we check for various hypotheses such as constant returns
to scale,
weak separability with linear separability restrictions, and Cobb-Douglas
hypothesis.
We find that log of likelihood ratios are 54.34, 70.70, and 100.14 for
constant returns
to scale, weak separability, and Cobb-Douglas hypotheses respectively.
While, the
0.01 critical values of Chi-square are 13.24, 11.34 and 16.81 for constant
returns to
scale, weak separability, and Cobb-Douglas hypotheses respectively.
Thus, we reject
the null hypotheses, and we conclude that constant returns to scale, value
added
specification, and Cobb-Douglas hypothesis are not satisfied with
Jordanian
manufacturing data.
Table 1 shows IZEF estimates of three-input Jordanian manufacturing
symmetric translog production function. The raw moment R-square for
capital and
labor cost share are 0.87, and 0.94 respectively. The estimates of translog
production
and cost functions, under different hypotheses, have been introduced in
Table 4 and 5
in the appendix.
Now, let us check for positivity and concavity of the function. For
positivity,
we find that fitted cost share of capital, labor and materials are all positive
at each data
point. Next, we check for concavity of production function, and we find
that Hessian
matrix, based on IZEF parameter estimates, is negative semidefinite
(Berndt and
Wood 1975). Thus, we can say that production function is well behaved
for
manufacturing industrial data of 2002.
In measuring substitutability of factor inputs, we compute AES by using
(3).
In addition, we calculate factor cross price elasticities by using (4). [3]
Where we use
fitted cost share of inputs and estimates of symmetric translog cost
function in our
estimations. However, our estimate of AES and _ij and _ji are presented in
Table 2 and
coefficient of variation of 0.139 (Norsworthy and Malmquist 1983). So, _KL
is close
6
to be constant but it is significantly different from one. This result is
similar to that of
Berndt-Wood, where the estimated _KL are rather stable time period of
study 1947-
71. We also find that factor input cross price elasticities, _KL and _LK , are
about 0.17
and 0.25 respectively.
(iii) Capital and material are slightly substitutive. Where, average _KM is
0.82 with
variance and coefficient of variation equal to 0.001 and 0.035
respectively. Factor
input cross price elasticities _KM and _MK are 0.53 and 0.17 respectively. (iv)
The
average of _LM is equal to 0.70 with variance of 0.009 and coefficient of
variation
equals to 0.139. The factor input price elasticities, _ LM and _ ML , are 0.45
and
0.10 respectively. (v) It is clear, that _KM is less variant than _KL and _LM .
(vi)
_KM is about two thirds of _KL . _LM is about eighty five percent of _KM and
about
fifty five percent of _LM . However, in Table 6, we introduce only thirty
estimates of
AES out of seventy seven estimates. In Table 7, we also introduce
estimated cross
price elasticities for selected manufacturing industries.
IV. SUMMARY AND CONCLUSIONS
Our objective has been to estimate a translog production function and
show its
internal structure for four digits Jordanian manufacturing industries in a
point of time.
We estimate parameters that represent the average cost share of inputs
and the
elasticities of cost share of input with respect inputs. In addition, we have
checked for
positivity, concavity, substitutability and separability of inputs. Our main
conclusions
are: (a) After testing for different hypotheses we find that three-input
symmetric
translog production function is an appropriate one for Jordanian
manufacturing
industry in 2002. (b) The Jordanian translog production function is well-
behaved. (c)
Capital-labor substitutability is larger than capital-materials and labor-
material. (d)
Allen partial elasticities of substitution are almost constant, but they are
significantly
different from one.
7
APPENDIX
Table 1. IZEF estimates of Jordanian Manufacturing
Translog Production Function*
(*) asymptotic t-ratio in parentheses
Table 2. IZEF Allen Elasticity of Substitution for
Jordanian Manufacturing Industry (*)
Each estimate represents an arithmetic average of 77 estimates of Elasticity of
(*)
substitution.
Table 3. IZEF Factor Input Price Elasticity for
Jordanian Manufacturing Industry (*)
(*)Each estimate represents an arithmetic average of 77 estimates of Factor Price
Elasticity.
Parameter Estimates Parameter Estimates
_K 0.1778
(14.498)
BKM -0.0237
_L 0.0810
(13.291)
BLL 0.0198
(2.883)
_M 0.7413
BLM -0.0253
BKK 0.0181
(2.812)
BMM 0.0490
BKL 0.0056
(1.018)
Elasticity Estimate Elasticity Estimate
_ KK -3.62 _ KL 1.25
_ LL -5.55 _ KM 0.82
_ MM -0.43 _ LM 0.70
Elasticity Estimates Elasticity Estimates
_ KK -0.70 _ LK 0.45
_ LL -0.70 _ LM 0.26
_ MM -0.27 _ MK 0.17
_ KL 0.17 _ ML 0.10
_ KM 0.53
8
Table 4. IZEF Translog Production Function Estimates-
Jordanian Manufacturing 2002 (*)
(*) asymptotic t-ratio in parentheses
Parameter
Unrestricted
CRS
Weak
Separability
Cobb-
Douglas
_K 0.1778 0.026 0.2062 0.20561
(14.498) (14.149) (12.375) (19.139)
_L 0.0810 0.1377 0.0966 0.1377
(13.291) (14.073) (10.600) (17.666)
_M 0.7413 0.6567 0.6972 0.6567
BKK 0.0182 0.0512 0.01575
(2.812) (3.683) (2.170)
BKL 0.0056 0.0112 -0.0319
(1.018) (1.990) (-6.246)
BKM -0.0237 -0.0624
BLL 0.0198 0.0264 0.0182
(2.883) (5.918) (2.825)
BLM -0.0253 -0.0376
BMM 0.0490 0.10000
R2
K
.
0.8679 0.8640 0.8175 0.8282
R2
L
.
0.9354 0.8766 0.8784 0.8042
LLF 224.85 197.68 189.50 172.78
9
Table 5. IZEF Translog Cost Function Estimates-
Jordanian Manufacturing 2002 (*)
(*) asymptotic t-ratio in parentheses
Parameter
Unrestricted
CRS
Weak
Separability
Cobb-
Douglas
_ K 0.1383 0.2056 0.1859 0.2056
(11.079) (8.957) (12.532) (19.139)
_ L 0.0808 0.1377 0.1022 0.1377
(8.025) (11.621) (10.628) (17.667)
_ M 0.7809 0.7413 0.7119 0.6576
_ KK 0.0705 0.0501 0.0072
(7.710) (5.320) (1.407)
_ KL 0.0223 0.0037 -0.0072
(3.377) (0.538) (-1.407)
_ KM -0.0929 -0.0537
(-8.552) (-4.303)
_ LL 0.0170 -0.0033 0.0072
(1.439) (-0.255) (1.407)
_ LM -0.0393 -0.0033
(-2.974) (-0.027)
_ MM -0.0971 0.0496
_ YK 0.0827 0.0602 -0.0090
(7.054) (4.082) (-1.416)
_ YL
0.0144 -0.0106 -0.0248
(1.120) (-0.705) (-6.069)
_ YM -0.071 0.0496
R2
K
0.9140 0.8755 0.8428 0.8282
R2
L
.
0.8776 0.351 0.8623 0.8042
LLF 215.88 192.17 189.159 172.77
10
Table 6. IZEF Estimated Elasticities of Substitution Translog Cost Function
for
Thirty Selected Jordanian Manufacturing 2002
Economic Activity ISIC _KK _LL _MM _KL _ K M _ LM
Production, Processing, and Preserving of 1511 -3.70 -7.51 -0.30 0.83 0.62
Meat and Meat Product
Manufacturing of Soft Drink; Production of 1554 -2.86 -5.25 -0.48 1.19 0.84 0.70
Mineral Water
Preparation and Spinning of Textile Fibber ; 1711 -3.59 -6.55 -0.34 1.29 0.82 0.67
Weaving of Textiles
Manufacturing of Made-up Textile Articles, 1721 -4.40 -3.50 -0.46 1.19 0.77 0.80
Manufacturing of Wearing Apparel, Except 1810 -3.16 -3.51 -0.58 1.14 0.81 0.78
Fur Apparel
Manufacturing of Luggage, Handbags and
the Like, Saddler and Harness
1912 -3.26 -2.59 -0.72 1.11 0.79 0.81
Manufacturing of Footwear 1 9 2 0 -2.29 -3.95 -0.57 1.15 0.83 0.76
S Sawmilling and Planning of Wood 2 0 10 -3.84 -4.63 -0.41 1.22 0.80 0.75
Manufacturing of Wooden Containers 2023 -2.86 -3.24 -0.67 1.12 0.82 0.78
Manufacturing of Pulp, Paper and
Paperboard
2101 -3.62 -6.01 -0.36 1.27 0.82 0.69
Manufacturing of Publishing of Newspapers 2212 -3.25 -4.75 -0.46 1.19 0.82 0.74
Journals and Periodicals
Manufacturing of Refined Petroleum Product 2 3 2 0 -9.91 -7.76 0.07 2.04 0.65 0.72
Manufacturing of Basic Chemicals, except 2411 -2.70 -7.54 -0.41 1.27 0.85 0.59
Fertilizers and Nitrogen Compounds
Manufacturing of Fertilizer and Nitrogen 2412 -6.19 -6.55 -0.08 1.49 0.77 0.73
Compounds
Manufacturing of Plastic in Primary Forms 2413 -3.02 -5.75 -0.43 1.22 0.84 0.69
and of Synthetic Rubber
Manufacturing of Rubber Tires and Tubes; 2511 -3.29 -4.94 -0.44 1.20 0.82 0.73
Retreating and Rebuilding of Rubber Tiers
Manufacturing of Plastic Products 2520 -3.85 -6.54 -0.32 1.31 0.82 0.67
Manufacturing of Glass and Glass Products 2610 -2.91 -4.86 -0.50 1.18 0.83 0.72
Manufacturing of Non-structural and 2691 -2.76 -4.26 -0.57 1.15 0.83 0.74
Non- refractory Ceramic Ware
Manufacturing of Structural Non-refractory
Clay and Ceramic Product
2693 -2.88 -4.39 -0.54 1.16 0.83 0.74
Manufacturing of basic iron and steel 2710 -4.09 -8.97 -0.25 1.48 0.82 0.55
Manufacturing of Structural Metal Products 2811 -3.93 -4.10 -0.44 1.20 0.80 0.77
Treatment and Coating of Metals, General 2892 -2.23 -2.91 -0.89 1.09 0.83 0.77
Mechanical Engineering on a Fee or
Contract Basis
Manufacturing of Agricultural and Forestry 2921 -6.36 -7.90 -0.18 1.63 0.74 0.64
Machineries
Manufacturing of Domestic Appliances, 2930 -3.57 -5.43 -0.39 1.24 0.82 0.72
Manufacturing of Electronic Motors 3110 -3.62 -6.36 -0.34 1.28 0.82 0.67
Manufacturing of Medical and Surgical 3311 -2.70 -4.08 -0.59 1.14 0.83 0.75
Equipments and Orthopedic Appliances
Building and Repairing of Pleasure and 3512 -3.15 -4.01 -0.53 1.16 0.82 0.76
Sporting Boats
Manufacturing of Furniture 3610 -3.53 -3.99 -0.49 1.18 0.81 0.77
Manufacturing of Jewelry and Related
Articles
3691 -4.48 -4.95 -0.34 1.27 0.79 0.75
11
Table 7. IZEF Estimated Factor Input Cross Price Elasticities of Demand
Translog
Cost Function for Thirty Selected Jordanian Manufacturing 2002
Economic Activity ISIC _ KL _ KM _ LM _ LK _ M K _ ML
Production, Processing, and Preserving of 1511 0.12 0.59 0.44 0.25 0.16 0.06
Meat and Meat Product
Manufacturing of Soft Drink; Production of 1554 0.16 0.52 0.44 0.28 0.20 0.10
Mineral Water
Preparation and Spinning of Textile Fibber ; 1711 0.14 0.57 0.46 0.25 0.16 0.07
Weaving of Textiles
Manufacturing of Made-up Textile Articles, 1721 0.24 0.49 0.51 0.19 0.13 0.16
Manufacturing of Wearing Apparel, Except Fur 1810 0.23 0.47 0.45 0.25 0.18 0.16
Apparel
Manufacturing of Luggage, Handbags and the
Like,
1912 0.28 0.42 0.43 0.24 0.17 0.21
Saddles and Harness
Manufacturing of Footwear 1920 0.20 0.48 0.44 0.27 0.20 0.14
S Sawmilling and Planning of Wood 2010 0.19 0.53 0.49 0.22 0.15 0.12
Manufacturing of Wooden Containers 2023 0.24 0.45 0.43 0.27 0.20 0.17
Manufacturing of Pulp, Paper and Paperboard 2101 0.15 0.56 0.47 0.24 0.16 0.08
Manufacturing of Publishing of Newspapers 2212 0.18 0.52 0.47 0.25 0.18 0.11
Journals and Periodicals
Manufacturing of Refined Petroleum Product 2320 0.17 0.67 0.74 0.13 0.04 0.07
Manufacturing of Basic Chemicals, Except 2411 0.12 0.56 0.38 0.31 0.21 0.05
Fertilizers and Nitrogen Compounds
Manufacturing of Fertilizer and Nitrogen 2412 0.16 0.67 0.64 0.17 0.09 0.08
Compounds
Manufacturing of Plastic in Primary Forms 2413 0.15 0.54 0.44 0.27 0.19 0.09
and of Synthetic Rubber
Manufacturing of Rubber Tires and Tubes; 2511 0.17 0.53 0.47 0.25 0.18 0.11
Retreating and Rebuilding of Rubber Tiers
Manufacturing of Plastic Products 2520 0.14 0.58 0.47 0.24 0.15 0.07
Manufacturing of Glass and Glass Products 2610 0.17 0.51 0.44 0.27 0.20 0.11
Manufacturing of Non-structural and 2691 0.19 0.49 0.43 0.28 0.21 0.13
Non- refractory Ceramic Ware
Manufacturing of Structural Non-refractory Clay 2693 0.19 0.50 0.44 0.27 0.20 0.12
and Ceramic Product
Manufacturing of Basic Iron and Steel 2710 0.11 0.61 0.41 0.25 0.15 0.04
Manufacturing of Structural Metal Products 2811 0.21 0.51 0.50 0.21 0.15 0.14
Treatment and Coating of Metals, General 2892 0.25 0.39 0.37 0.31 0.24 0.18
Mechanical Engineering on a Fee or
Contract Basis
Manufacturing of Agricultural and Forestry 2921 0.14 0.59 0.51 0.18 0.09 0.06
Machineries
Manufacturing of Domestic Appliances, 2930 0.16 0.55 0.48 0.24 0.16 0.10
Manufacturing of Electronic Motors 3110 0.14 0.57 0.47 0.25 0.16 0.08
Manufacturing of Medical and Surgical 3311 0.20 0.48 0.43 0.28 0.21 0.13
Equipments and Orthopedic Appliances
Building and Repairing of Pleasure and 3512 0.20 0.49 0.46 0.25 0.18 0.14
Sporting Boats
Manufacturing of Furniture 3610 0.21 0.50 0.48 0.23 0.16 0.14
Manufacturing of Jewelry and Related Articles 3691 0.18 0.55 0.52 0.20 0.13 0.11
12
ENDNOTES
[1] There are two alternative tests that can be used for testing
hypotheses, Wald
test, and Lagrangian Multiplier test
[2] The translog cost function is:
Ln TC= _o + _Y Ln Y+1/2_YY Ln Y2+ __i Ln Pi +1/2_ _ _ij LnPi Ln Pj. + _
_Yi LnY Ln Pj
and by differentiating the function we get:
_ Ln TC / _ Ln PK = _ K + _KK Ln PK + _KL Ln PL+ _KM Ln PM + _KY Ln
Y
_ Ln TC / _ Ln PL = _ L + _LK Ln PK + _ LL LnPL+ _ LM Ln PM + _LY
Ln Y
_ Ln TC / _ Ln PM = _ M + _MK Ln PK + _ ML LnPL+ _ MM Ln PM + _MY
Ln Y
[3] Cross price elasticities of factor inputs can, also, be estimated as
follows:
_ij=Sj_ij
13