i Trade Liberalization, Poverty and Welfare in Pakistan Inaugural Dissertation submitted as the requirement for the degree of PhD International Development Studies (IDS) at the Institute of Development Research and Development Policy Ruhr-University Bochum Submitted by Naveed Ahmed Shaikh Bochum 2011
206
Embed
Trade liberalization, poverty and welfare in Pakistan · i Trade Liberalization, Poverty and Welfare in Pakistan Inaugural Dissertation submitted as the requirement for the degree
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
i
Trade Liberalization, Poverty and Welfare in Pakistan
Inaugural Dissertation submitted
as the requirement
for the degree of
PhD International Development Studies (IDS)
at the
Institute of Development Research
and Development Policy
Ruhr-University Bochum
Submitted by
Naveed Ahmed Shaikh
Bochum 2011
ii
Table of Contents
List of Figures v
List of Tables vi
Abbreviations viii
Words of Thanks ix
1 Introduction 1
2 International Trade-Labour Income Inequality, Prices, and Poverty 8
2.3 Stolper-Samuelson (S-S) Theory and Heckscher-Ohlin (H-O) Model 13
2.3.1 Assumptions and Implications of the Chosen Approach 14 2.3.2 Implications of the Model 19
2.4 Evidence from the Literature 20 2.4.1 Evidence from Latin America 21 2.4.2 Evidence from Asia 23
2.5 Discussion on Empirical Evidence 25
3 Choice of Methodological Technique 27
3.1 Computable General Equilibrium (CGE) Analysis 28
3.2 Partial Equilibrium Analysis 33
3.3 Micro Macro (Simulation) Models 36
3.4 Choice of an appropriate Modeling Technique 38
4 Trade Liberalization, Prices of Traded and Nontraded Goods, Households’ Labour Income, Welfare, and Poverty 41
4.1 Interrelations between International Trade, Domestic Prices, and Factor Prices 42
4.1.1 Domestic Prices of Traded Goods 42
iii
4.1.2 Domestic Prices of Nontraded Goods 44 4.1.3 Households’ Labour Income 45
4.2 Trade Liberalization, Household Demand, and Welfare Effects 58
4.2.1 Household Expenditure 58 4.2.2 Change in Marshallian Consumers’ Surplus 63 4.2.3 Change in Hicksian Compensating Variation 65 4.2.4 Change in the Households’ Labour Income 67 4.2.5 Change in Poorest Households’ Welfare 69
5 Statistical Results and Interpretation 72
5.1 Data 73 5.1.1 Import Tariff and International and Domestic Prices 73 5.1.3 Domestic Prices and Labour Income 78 5.1.4 Exclusion of Goods from the Model 79
5.2 Regression Analysis 80
5.3 Change in the Domestic Prices of Traded and Nontraded Goods and Households’ Demand 82
5.3.1 Change in the Domestic Prices of Traded Goods 82 5.3.2 Change in the Domestic Prices of Nontraded Goods 84 5.3.3 Household Demand Equations and the Change in Demand
for the Selected Traded and Nontraded goods 87 5.3.3.1 Estimated Marshallian Demand 89 5.3.3.2 Estimated Hicksian Demand 93
5.4 Trade Liberalization, Household Welfare, Poorest Household Welfare and Labour incomes 96
5.4.1 Welfare Measuring Approaches 96 5.4.2 Marshallian Consumer Surplus (MCS) 99 5.4.3 Hicksian Compensating Variation (HCV) 102 5.4.4 Poorest Households’ Demand and Welfare (MCS) 107 5.5 Labour Incomes and the Domestic Prices of the Selected
Goods 111 5.6 Total (Price and Labour income) Effect on Household
Welfare 119
5.7 Discussion on the Statistical Results 124
5.8 Limitations of the Study 129
6 Summary and Policy Recommendations 131
iv
Bibliography 136
Appendix 146
A1 Data and Other Descriptive Tables 146
A2 Estimated Demand Equations 154
A3 Detailed tables on empirical estimations of demands and domestic prices including tariff 156
A4 Detailed Tables on Import Tariff and Calculation of Commodity-wise Tariff Following the Tariff in the General Economy 165
A5 Detailed tables on welfare loss due to selective protectionist trade policy in selected traded goods 172
A6 Domestic Prices, Production and Trade of selected goods for 1970-2005 176
v
List of Figures
Fig. 1: Customs revenue in percent of imports (C.I.F) in Pakistan (1992-2005) 2
Fig. 2: PCI-Trend in Pakistan (1992-2005) 2
Fig. 3: Commodity-groups wise actual tariff in PKR per ton calculated by dividing the collected tariff revenue in PKR by the import quantities in tons (3-11) 4
Fig. 4: Relative commodity prices determine the wage-rent ratio 17
Fig. 5: Factor-Price diagram with unit cost curves (left) with no Factor Intensity Reversals and (right) with Factor Intensity Reversals 49
Fig. 6: Factor-Price Diagram reflecting change in factor price (wage) due to change in one good's price while holding the prices of other goods constant. [Suranovic (2010)] 54
Fig. 7: A Linear case of 2x2x2 in Factor-Price Diagram 56
Fig. 8: Locating the lowest level of household expenditure to attain a utility level "u" 59
Fig. 9: The Hicksian and Marshallian demands for good X1 when its price P1 is falling 61
Fig. 10: Estimated and calculated domestic prices at average and real tariff rates for the available years 76
Fig. 11: Average percentage sector-wise employment in Pakistan from 1970-2005 (Various Labour Force Surveys) 117
Fig. 12: Average sector-wise employment of unskilled workers in Pakistan 1970-2005 (Various Labour Force Surveys) 118
vi
List of Tables
Table 1: 14-Year average domestic prices calculated at average actual and falling general import tariff rates (PKR per Ton) given average international prices and percent difference in both tariffs 83
Table 2: Empirical link between domestic price of electricity and the domestic prices of traded goods 85
Table 3: Empirical link between domestic price of firewood and the domestic prices of traded goods 85
Table 4: Average estimated domestic prices of nontraded goods at actual and at general falling tariff 87
Table 5: Percentage shares of monthly household expenditure for commodities from various sectors 88
Table 6: Estimated linear and natural log linear Marshallian demand equations 91
Table 7: Average Estimated Marshallian demand quantities in tons at actual tariff and at the falling general tariff 92
Table 8: Average Estimated Hicksian demand quantities in tons at actual tariff and at the falling general tariff 94
Table 9: Total and per household change in MCS due to the selective protectionist trade policy in the selected commodities 1992-2005 100
Table 10: Yearly total and average change in welfare under MCS to the ordinary Pakistani household 101
Table 11: Total and per household change in HCV due to the selective protectionist trade policy in the selected commodities 1992-2005 103
Table 12: Yearly total change in HCV and change in HCV per average Pakistani household 104
Table 13: Comparison of Total HCV and MCS in PKR 105
vii
Table 14: 14 year total sum of the poorest household demand quantities (KGs) of the selected goods at actual tariff and when the tariff follows the general falling tariff 108
Table 15: Single poorest household’s change in MCS due to the selective protectionist trade policy 1992-2005 109
Table 16: Year-wise change in MCS of a single poorest household in all selected traded and nontraded goods due to protectionist trade policy 110
Table 17: Empirical link between labour incomes in agriculture and the domestic prices 112
Table 18: Empirical link between labour incomes in manufacturing and estimated domestic prices 113
Table 19: Estimated agriculture labour income with domestic prices at actual tariff and at domestic prices if the general falling tariff is applied 113
Table 20: Estimated manufacturing labour income with domestic prices at actual tariff and at domestic prices if the general falling tariff is applied 115
Table 21: Inter-sector labour income correlation 118
Table 22: Net change in an ordinary household’s labour income due to the protection 120
Table 23: Total yearly loss to an ordinary Pakistani household due to the selective protectionist policy 121
Table 24: Total welfare loss to the poorest Pakistani household due to the protectionist policy 122
viii
Abbreviations
CGE Computable General Equilibrium
C.I.F Cost Insurance and Freight
CUM Cubic Meter
D-W Durbin Watson
FAO Food and Agriculture organization
FBS Federal Bureau of Statistics
FGT Foster Greer Thorbecke indicators
Fig. Figure
GDP Gross Domestic Product
GTAP Global Trade Analysis project
H-O Heckscher-Ohlin
HCV Hicksian Compensating Variations
HEC Higher Education Commission
IDS International Development Studies
ILO International Labour Organization
MCS Marshallian Consumer Surplus
OLS Ordinary Least Squares
PCI Per Capita Income
PKR Pakistani Rupee
SAM Social Accounting Matrix
S-S Stolper-Samuelson
ix
Words of Thanks
I am pleased to acknowledge the kind assistance, scholarly guidance, and
help of many people and institutions by expressing my heartfelt gratitude
and thankfulness toward them.
The first amongst the individuals, who supported, helped, and inspired me
throughout my doctoral thesis is my PhD supervisor, Prof. Dr. Wilhelm
Löwenstein. Indeed it was his continuous support and active supervision at
every stage of this piece of writing that enabled me to accomplish such
advanced research work. I am even more indebted to him especially for
his kind consideration for sparing extra and spontaneous spans of time
during the final weeks before submission of the thesis. Secondly, I pay
special thanks to Prof. Em. Dr. Dieter Bender for being co-supervisor of
my thesis. Thirdly, I would take the opportunity to acknowledge the
enlightening discussion(s) with Dr. Tobias Bidlingmeier, one of my
colleagues at Institute of Development Research and Development Policy,
Ruhr-University Bochum, working on a similar topic, on issues related to
model building and variable identification at early stages of my work.
In Pakistan my sincere thanks go to Dr. Ghulam Murtaza Khuhro, Deputy
Commissioner, Income Tax, Karachi for his support in collecting
statistical books, annual reports, and other published materials from the
library of Statistical Division of Pakistan, Karachi Branch. My special
thanks go to Mr. Bashir Ahmed Zia, Chief Librarian who took special
efforts and invested time in getting the bundles of data Chapters from
various annual surveys and reports photocopied and sending me in
Germany at State Bank’s expenses on my request.
Amongst institutions, I express my earnest thankfulness to Higher
Education Commission of Pakistan for its financial assistance for five
x
years that facilitated me to earn my PhD in Germany. Secondly, my thanks
proceed for DAAD (German Academic Exchange Service) for smoothing
my placement as a PhD student at the Institute of Development Research
and Development Policy, Ruhr-University Bochum and my stay in
Germany throughout the study period. I admit that without DAAD
support, stay in Germany would not have been as pleasant and
comfortable as it was.
The present PhD International Development Studies (IDS) program
included a three-month field survey for data collection in Pakistan. Being
a member of the Ruhr University Research School1 at Ruhr University
Bochum, my field survey trip to Pakistan was funded from the annual
allowance of the Ruhr University Research School. Further, different
workshops and seminars offered by Ruhr Research School played a great
role in creating a serious research environment amongst PhD scholars
from the variety of disciplines. The workshop I found most useful while
writing my PhD thesis was on becoming a better academic writer. For all
this I extend my heartfelt gratitude for Dr. Ursula Justus, Counseling (PhD
Planning and Funding Opportunities) and Ms. Maria Sprung, Assistant
from Ruhr Research School, for their support.
I also thank the staff of State Bank of Pakistan, Central Directorate,
Central Library administration for absolute cooperation in accessing the
books, journals, and annual reports during my visit there.
Last but not least, I express my thanks for the staff and colleagues at
Institute of Development Research and Development Policy for extending
a cooperative and helping hand whenever I approached them.
At the end, I would like to mention the support and the help extended from
my loving wife, Shamshad Naveed. Despite that she was student of Master
1 http://www.research-school.rub.de/about_us.html
xi
of Science in Computer Engineering at Duisburg-Essen University,
Duisburg Campus, she took care of me, our home and our child during my
busy schedules. I also want to mention the prayers and motivation of my
father back in Pakistan which have always encouraged me and have been a
source of resilience in my life.
Naveed Ahmed Shaikh,
IEE, Bochum, 2011
1
1 Introduction
Recent decades have observed rapid expansion in the monetary worth of
the world economy. With the inception of the era of economic
globalization since the last two decades of the 20th century, countries have
drawn closer to each other for more trade integration and economic
cohesion. Though the swollen volume of global trade may have brought
fortunes for the world economy and for some individual countries2
(Example: export-oriented growth in East Asian countries after
liberalizing trade during the 1960s and 1970s) nevertheless it has raised
several matters of serious concern regarding the impact of globally freer
trade on the poverty situation, with special emphasis on developing
countries. Some of the crucial questions confronting researchers in the
fields of development studies and international trade are: Does enhanced
volume of global trade help control the global poverty rate? Or do open
developing countries outperform the closed ones in attaining the national
poverty targets and pursuing the well being for their populations? Do poor
masses in developing countries benefit from the international trade and
lose from protectionist policies? To reach some reasonable conclusions
regarding trade-poverty and wage inequality links (Sections 2.1 and 2.2) in
light of the above mentioned questions, an intensive literature survey is
conducted and presented in section 2.4. Existing literature on the
experience of several developing countries with liberalizing trade regimes
provides an inconclusive blend of arguments with findings for and against
the liberal policies. The case of Latin American and Asian countries’
liberal trade policies is discussed in sections 2.4.1 and 2.4.2.
2 Ahmed, J. (2001) has found a strong two way causality between exports and income growth. The discussion on issues related with trade-growth causality is provided in section 2.1.
%
Pakistan being a developing country has followed an impressive overall
liberal trade policy by slashing the general customs tariff in percent of
imports3 from 19.88% in 1992 to 9.81% in 2005 (Figure 1). Further, the
Per Capita Income (PCI) during the same period almost doubled from US
$449.61 in 1992 to US $833.04 in 20054 (Figure 2).
Fig. 1: Customs revenue in percent of imports (C.I.F) in Pakist
Fig. 2: PCI-Trend in Pakistan (1992-2005) Source: Federal Bureau of Statistics (FBS), Pakistan
However, the customs revenue in percent of impor
related agricultural commodity groups and on fuels
an increasing trend indicating the adoption of a
3 calculated at Cost Insurance and Freight (C.I.F) 4 In Pakistani rupee terms the PCI increased by more than four 1992 to PKR 49841 in 2005.
Year Year
an (1992-2005)
Year
US $
t
s
t
2
s (cif) on most food-
and oils has followed
elective protectionist
imes from PKR 11249 in
3
policy (See Figure 3: Charts 1 to 9) during the same period. Out of nine
selected traded primary commodity groups the only commodity group i.e.
fruits, nuts and vegetables (Chart 1) depicts a clearly falling trend in the
tariff. Tea, coffee and spices group (Chart 2) has followed a falling trend
only after 2000. Except in the year 1997 when it jumped to PKR 45354.46
from PKR 23352.79, the tariff per ton has remained under PKR 30000.
The group seems to be liberalized except in the years when government
intervention pushed the tariff rate up to the unprecedented level. Two
commodity groups; milk, butter and cheese (Chart 3) and animal and
vegetable oil (Chart 4) depict a clear rising trend in the tariff. In case of
edible cereals and vegetables (Chart 5) the tariff rate remained low for the
period 1992-2000, however during 2001-2004 the group was protected
with high tariff rates. In tobacco sector (Chart 6) the tariff per ton has
remained as low as under PKR 1.5 million most of the time except two
years when it jumped to PKR 5.5 million in 1998 and PKR 17.8 Million in
2001. Therefore it may be presumed that tobacco sector is following the
liberal trade policy and rise of tariff in two years is not the part of the long
term trade policy but the isolated shocks. In case of fuels and oils (Chart
7) the rising trend in tariff revenue (thus tariff per ton) from 2001 to 2005
seems to reflect the increase in the oil price in international market during
the same time period. The tariff per ton in case of Sugar and confectionary
(Chart 8) remained under PKR 3000 except for the three years. It rose
dramatically in the years of 1998 to PKR 4940, to PKR 12674.27 in 1999
and to PKR 6866.90 in 2004 before slashing down to PKR 1416 in 2000.
In case of meat, fish and other preparations (Chart 9), the tariff per ton
during 1992-2005 has no clearly visible trend.
Fc
9
ig. 3: Commodity-groups wise actual tariff in ollected tariff revenue in PKR by the import qua
8
7
6
5
4
3
2
1
4
PKR per ton calculated by dividing the ntities in tons (1-9)
5
Precisely, one can conclude from the above Figure 3 (Charts 1 to 9) that
these commodity groups have gone through a selective protectionist
policy, in some cases commodity groups were protected with high tariff
rates for more than half of the time period of the study (1992-2005), in
contrast to relatively liberal trade policy in general in the economy with a
falling trend in the general tariff rate.
Although, on the other hand, the doubling of per capita income in parallel
to the 50% fall in general tariff rate during the same time period reflects a
successful trade and development strategy, nevertheless the rise in the
commodity-wise tariffs on selected commodity groups might have
defeated the overall gains of trade liberalization in the economy in general.
Therefore the present study attempts to compute the loss in the ordinary
and the poorest households’ welfare under a selective protectionist trade
policy in selected commodity groups against the relatively liberal trend in
the economy in general using Hicksian and Marshallian welfare measuring
approaches. More detailed discussion on the welfare measuring
approaches is presented in section 5.4.1. Further the gain and loss in the
labour income5 of agriculture and manufacturing workers under selective
protectionist policy against the general relatively liberalizing trend is
measured using Heckscher-Ohlin (H-O) and Stolper-Samuelson (S-S)
approach. The loss in the poorest households’ welfare is measured by
estimating the Marshallian Consumer Surplus (MCS) using poorest
households’ budget shares and the loss in agriculture labour income. The
discussion on H-O and S-S models will be presented in section 2.3.
Though the study is more about household analysis at a micro level,
nevertheless it also attempts to link domestic household welfare with the
trade reforms at the national level.
The reviewed literature on the liberalized trade effects on different sectors
of the economy in various developing countries contemplates the use of
5 Labour income includes wages and working hours
6
diverse methodologies varying across different dimensions, such as
whether the analysis is carried out for representative or actual households
or whether it is dynamic or static or using single- or multi-regional
statistics and whether it uses partial or general equilibrium approaches. In
the present study, the Partial Equilibrium approach, with theoretical
support of the General Equilibrium Framework in background, is preferred
over other techniques for the purposes of simplicity of the model,
exactness and precision in estimating the impact of policy change and the
availability of data in a developing country setting. The detailed account
of the choice of the appropriate methodological technique is presented in
Chapter 3. The interrelations between international trade, domestic prices
and wages in the light of theoretical findings are presented in Chapter 4.
The empirical results are presented in Chapter 5. The determination of the
domestic prices of selected traded and nontraded goods under selective
protectionist policy against a relatively liberalizing general economy is
presented in section 5.3.1 and 5.3.2. Without estimating, the domestic
prices of all traded goods are calculated by adding the tariff per ton to the
respective international prices. The adjusted domestic prices of traded
goods are used to estimate the domestic prices of two nontraded goods
(electricity and firewood) in the S-S setting in log-linear form. The change
in household welfare using MCS (see section 5.4.2) and HCV (see section
5.4.3) approaches is measured using the domestic prices under selective
protectionist policy and those calculated with falling general tariff under
relatively liberal policy in general economy. The Marshallian approach
quantifies the change in households’ welfare by measuring the change in
the households’ Consumer Surplus when the domestic prices of the
selected commodity groups under selective protectionist policy are
recalculated with the falling trend in the tariff in general. MCS can only be
used here under the restrictive assumption that a cardinal utility function
describes the Pakistani households’ preferences. Since the utility function
of the Pakistani households is unknown therefore HCV (Households’
7
willingness to pay) is estimated to know the accurate impact of the price
adjustment on Households’ Welfare using slutsky equation to exclude the
income effect from the total effect of the price changes that are induced by
alternative assumptions on the development of import tariffs. The welfare
loss to the poorest households is measured using poorest households
demand equations (calculated from their budget shares, yearly household
incomes and the domestic prices) in section 5.4.4. Additionally, the
expected variation in the labour income in agriculture and manufacturing
sectors is estimated in log-linear form assuming the commodity-wise tariff
had followed the falling trend in the general tariff (see section 5.5). Since
the selected goods are not sector-wise classified except agriculture and
manufacturing, the labour income and the price link is estimated only for
the above two sectors. The change in labour income in the construction
and wholesale and retail trade sectors is only speculated from the inter-
sector labour income correlation index.
Finally, the total sum of the two effects, i.e., the household welfare impact
(Hicksian) due to a change in domestic prices of traded and nontraded
goods and the impact on households’ income due to resulting change in
labour income is the total impact of trade reforms on households’ welfare
in Pakistan (see section 5.6). The total impact on the poorest households’
welfare is the sum total of the loss in their MCS added by the loss in the
agriculture labour income. The discussion on the empirical results and the
limitations of the study are presented in sections 5.7 and 5.8. The
summary and policy implications of the study are provided in the last
Chapter 6.
8
2 International Trade-Labour Income Inequality, Prices, and Poverty
The body of literature analyzed during the study is broadly classified into
two segments: one segment deals with the impact of trade on the country’s
national economic growth. From this channel, however, the poor can only
gain proportionately from enhanced growth, given that there are no
income distributional transformations after trade reforms. The second
segment offers analysis of the trade-poverty link via changes in domestic
prices of traded and nontraded goods and wages. Since trade liberalization
affects household welfare by altering domestic prices of traded and
nontraded goods and wages of workers in various sectors of the economy,
the present study bases trade patterns of Pakistan with rest of the world on
the H-O model, and the link of trade with domestic prices and labour
incomes is determined from the S-S theorem. The present Chapter is
divided into two parts. The first part describes the theoretical approach of
the S-S theory and the H-O model. The second part discusses existing
evidence on the impact of liberalized trade policy on wages, prices, and
economic growth. Prior to making any proceedings with the subject
matter, it seems logical to first look at the transmission channels through
which trade affects poverty. For a detailed discussion on the links between
global trade and poverty see Harrison, A. and McMillan, M. (2007).
The (indirect channel) link between trade, growth and poverty is quite well
established in the literature. Several studies [such as Dollar and Kraay
(2001) and Sachs et. al. (1995)] claim that trade is good for the economic
growth of developing countries and that open economies outperform
9
closed ones in achieving rapid economic growth. Esfahani (1991) has
concluded that export expansion resulting from trade reforms leads to the
availability of more imports, which spurs output in semi-industrialized
countries. The argument put forward to verify the former claim is based on
comparison between performance of Latin America and East Asia during
1965-19896 on three key variables- namely GDP growth rate, annual rate
of growth in the manufacturing sector, and growth in national exports.
Latin American countries followed the dictates of import substitution
policy and showed poor performance in contrast to rapidly growing East
Asian countries implementing an outward oriented strategy. In addition,
some studies found a predominantly positive relationship between exports
and economic growth by employing cross country regression analysis.7
On the other hand, there is evidence that global economic growth along
with the spread of technological innovation and the substantial diminution
of the barriers to international trade are regarded as the raison d’être of the
rise in the volume of global trade to historically unprecedented levels.
Rodriguez, F. and Rodrik, D. (1999), Ravallion (2004), Agenore (2002)
and others tend to contest the generalized mainstream view about the
causal association between trade and growth. They recommend
methodological improvements in empirical strategies and supplementary
social protection policies to ripen the fruits of trade. The problems with
using export volumes in these regressions are the endogeneity of trade and
the undetermined exports-economic growth causality, since trade is not
exogenous but rather is influenced by various other factors, especially
economic growth.
The issue of endogeneity and causality of trade and growth is tackled in
Frankel and Romer (1999) by introducing the geographical factor as the
instrument variable for trade. It assumes that geographical distance
6 World Bank (1989, 1990). Also cited in Edwards (1993). 7 See Edwards (1993) for review of the related literature.
10
influences trade volume but is independent of income (growth). Their
results demonstrate a positive but statistically weak link between trade and
income and therefore cannot be delivered as a rigorous proof8.
Rodriguez, F. and Rodrik, D. (1999) further took a skeptical approach
toward the causal association between trade reforms and economic growth
and showed that geography can influence other important factors such as
institutions besides trade. Therefore it cannot be concluded that trade
causes rise in income. However, they found a slightly negative
relationship between import duty and economic growth rate using data
from 124 countries from 1975-1994. Even though studies applying
geographical distances to predict trade shares obtain rigorous results, still
one cannot definitively say that trade causes a rise in income.
Ravallion (2004), by using cross country comparisons and aggregate time
series data (macro lens) and household-level data combined with structural
modeling of the impact of rising trade volumes of 75 countries (micro
lens), also cast doubt on the impact of trade reforms on growth and
poverty devoid of well-designed social protection policies. Agenore
(2002) suggests a “transition period” after assimilation of technological
transfer by developing countries, when globalization may only have a
limited effect on poverty and growth. Quite in line with the above study is
the study of Glenn W. Harrison, Thomas F. Rutherford, and David G. Tarr
(2001). They illustrate that trade reforms may result in aggregate welfare
gains for the households in Turkey; however, it is possible that the poorest
households may lose because of adverse distributional consequences of
trade reform. The authors, though ambiguously, suggested direct
compensation to the poor or implementation of trade reforms in a limited
way to provide space to the poorest households. However, this can only be
done when the sources of the change in inequality are decomposed. Using
8 p 394f
11
Shorrocks’ (1982) decomposition approach9, they identified that the
principal reason for the poor losing is the fall in the wage of production
labor in the manufacturing sector.
Moreover it is relevant to discuss Khan (1998) regarding trade
liberalization experience in Pakistan. Khan (1998) argued that growth in
all exports and growth in manufactured exports in particular are important
for economic growth in Pakistan. Trade openness supports the export-
oriented production base of the country and facilitates growth prospects. If
the findings in Dollar and Kraay (2001) are arbitrarily accepted on
statistical and technical grounds that growth is good for the poor, then
indirectly it can be predicted that trade will have a pro-poor impact in the
case of Pakistan.
2.2 Trade-Price-Wage-Poverty Nexus
The link is based on the theory of comparative advantage in trade. The H-
O theorem depicts that the country’s comparative advantage in trade is
determined from the endowment of its production factors. Countries
endowed with abundant labor have an advantage in cheap labor costs of
producing goods, and countries endowed with abundant capital have an
advantage in producing capital-intensive goods at low production costs.
Thus, labor-abundant developing countries produce and export labor-
intensive products, and capital-abundant countries produce and export
capital-intensive products. The adjustment in the relative domestic prices
of traded and nontraded goods in the trading countries are determined on
the lines of the S-S theorem. This theorem proposes an adjustment in the
relative domestic price of a good, which leads to adjustment in the return
to the factor that is used most intensively in the production of the good.
9 Applying inequality decomposition rules based on variance and Gini-coefficients
12
Previous literature on the link between international trade liberalization
and poverty through labour incomes and domestic prices provides a mix
reaction in different developing countries.
Siddiqui, R. and Kemal, A.R. (2002), working on a link between trade
liberalization, and poverty in case of fall (or no fall) in the foreign
remittances. Using Computable General Equilibrium framework they
found that the rise in the poverty after implementation of liberal trade
policy during 1980s was a result of fall in the foreign remittances and the
tariff reduction indeed had resulted in a fall in poverty in both the rural
and urban areas of Pakistan. In terms of welfare, all households appear to
gain. The results show that the gain in welfare is larger for urban
households than for rural households. In addition, the predicted reduction
in poverty is larger (in percentage) in urban households than in rural
households.
Bleaney (1993) concluded that a global policy shift in the developing
world toward greater outward orientation may depress prices of
agricultural commodities and hence worsen the terms of trade of
developing countries. Further they suggested that the direct income effects
of this may likely be small, the indirect effects working through a
tightening of balance-of-payments constraints could be of considerable
significance and may entirely offset the expected gains from trade
liberalization.
The results found in Minot, N. and Blauch, B. (2002) indicate that export
liberalization would raise the price of rice and hurt the urban poor and
rice-deficit households in Vietnam. At the same time, gains in the rural
sector, particularly among farmers in the delta regions, outweigh these
effects, resulting in a slight reduction in overall poverty and an increase in
household and national income. Kim, K.S. and Vorasopontaviporn, P.
(1989) show that, for Thailand, more trade is likely to increase the demand
for low-labour income agricultural labor. Saggay, A. et al. (2006) found a
13
negative effect from import competition on domestic prices in case of
Tunisian manufacturing industries. Yang, Y. Y. and Hwang, M. (1999)
found a restraining effect of import competition on domestic prices in
Korea.
2.3 Stolper-Samuelson (S-S) Theory and Heckscher-Ohlin (H-O) Model
The theories of international trade and integration in the world economy
are as old as the Theory of Absolute Advantage given by the neoclassical
economist Adam Smith in 1776. In The Wealth of Nations, he argued that
“the invisible hand” of the market mechanism, rather than government
policy, should determine what a country imports and exports. Later on two
theories emerged from Smith’s Theory of Absolute Advantage. First,
David Ricardo’s Theory of Comparative Advantage came in 1817. The
principle of comparative advantage states that a country should specialize
in producing and exporting those products in which it has a comparative or
relative cost advantage compared to other countries and should import
those goods in which it has a comparative disadvantage. It is argued
further that the greater benefit for all trading partners would accrue out of
such specialization. Second, the theory previously called factor
proportions theorem was developed by two Swedish economists, Eli
Heckscher and Bertil Ohlin, in 1933. This theory later became popular as
Heckscher-Ohlin (H-O) theory. Since trade liberalization affects
household welfare by altering the domestic prices of traded and nontraded
goods and labour incomes by affecting wages of workers in various
sectors of the economy, the study in hand incorporates specifications of
two theorems as the theoretical background of the study: H-O Trade
Theorem and S-S Theorem.
14
The H-O theorem rationalizes the idea of trade relations of a developing
country with the rest of the world, and the S-S Theorem describes the
association of movement of the relative prices of commodities with the
movement of the relative prices of factors (wage and capital rent) in a
small open economy.10 The assumptions of the H-O model follow in the
next subsection.
2.3.1 Assumptions and Implications of the Chosen Approach
The model is also known as the 2x2x2 model since it preliminarily
assumes the world with two countries (A and B), two goods (X and Y),
and two factors of production (K and L). The total amount of labor and
capital used in production is limited to the endowment of the country.
Thus the labor constraint for a country is LX + LY ≤ L. Here LX and LY are
the quantities of labor used in production of X and Y goods, respectively.
L represents the labor endowment of the country. Capital constraint is KX
+ KY ≤ K. Here KX and KY are the quantities of capital used in the
production of two goods X and Y, respectively. K represents the capital
endowment of the country. Full employment of capital and labor implies
that the expression would hold with equality in both of the above
inequalities.
Thus, the trading countries only differ in their endowments of capital and
labor.
Two Goods
X and Y are the only goods produced by the two countries. It is assumed
that X is labor-intensive and Y is capital-intensive.
10 Assumptions of Constant Returns to Scale, Perfect Competition, and Equality of number of Factors to the number of products apply.
15
Two Factors
Two factors of production, labor and capital, are used to produce the
assumed two goods. Both labor and capital are homogeneous. Thus there
is only one type of labor and one type of capital. It is also assumed that
labor and capital are freely mobile across industries within the country but
immobile across countries.
Factor Constraints
A country is capital-abundant relative to another country if it has more
capital endowment per labor endowment than the other country. Thus in
this model A being the developed country is capital abundant relative to B
if:
K/LA > K/LB or L/KA < L/KB
Here K/LA is the capital-labor ratio in country A so it is a capital abundant
country as it is using more capital per unit of labor, and K/LB is the
capital-labor ratio in country B so B is labor-abundant country as it is
using more labor per unit of capital.
The original model of Heckscher and Ohlin assumed that the only
difference between countries is of the endowments of labor and capital.
The results of the seminal work by Heckscher and Ohlin have been the
formulation of certain conclusions arising from the assumptions inherent
in the model. The following description about model, assumptions, and
factor constraints are based on the textbooks on Internal Economics
[Appleyard, Field, and Cobb (2006) and Case, Karl, E., and Fair, Ray C.
(1999)]. These conclusions are better known as various theorems of the
model, which are given as follows:
1. H-O Theorem: One country’s comparative advantage in trade is
determined by its relative endowments of production factors.
16
Countries enjoy a comparative advantage in trading those goods
which use a relatively abundant factor of production more
intensively. This is because the profitability of goods is established
by the incurring input costs.
2. Factor Price Equalization Theorem: Relative prices for two
identical factors of production between two commodities will
equal each other because of trade and competition.
3. The S-S Theorem: A rise in the relative price of a good will lead
to a rise in the return to that factor used most intensively in the
production of the good, and conversely, to a fall in the return to the
other factor.
The H-O theorem predicts that a country will export the good as far as it is
relatively cheaper in its domestic production and import that which is
more expensive to produce domestically. The open trade-induced change
in the relative prices of goods in the domestic market affects the returns to
the employed factors of production. In autarky, the labor-intensive good is
cheaper in the labor-abundant country. In the case of free trade, the
relative prices of YK and XL equalize everywhere. Therefore the relative
price of YK (the capital-intensive good) rises in the capital-abundant
country, and the relative price of XL rises in the labor-abundant country.
This pushes the wage-rent ratio up in the labor-abundant country by
rewarding labor and punishing capital and lowers the wage-rent ratio in
the capital-abundant country by rewarding capital. Hence the model
suggests that countries will export the product that requires relatively more
of the abundant factor of production and import the good that requires
more of the scarce factor of production. In developing countries the use of
more unskilled labor increases demand for labor as mentioned earlier; as
the export sector expands due to liberalization so wages are likely to rise
relative to the rent to the capital. The determination of labour income-rent
ratio from information on relative prices of commodities is depicted in
Figure 4.
Fig. 4: Relative commodit
It is clear from Figure
the prices of factors o
intensive good) will p
labor-abundant count
X, (Px/Py)1, the relati
determined at (Kx/Lx)
When the relative pr
labour income to rent
income-rent ratio, the
and Y should also i
respectively. On the
don’t seem to be dire
them in one or all of t
• If nontraded
demand shift
consequently
goods lose, as
• If nontraded
demand for no
Wage-Rent Ratio
y prices determine the wage-rent ratio
4 that the relative prices of goods tr
f production. A rise in the relative pr
ush the labour income relative to cap
ry. Starting from the initial point of r
ve demand for capital in production
1, and in the production of good Y i
ice of X rises from (Px/Py)1 to (Px/P
must increase to (w/r)2. Now at the i
relative demand for capital in produc
ncrease correspondingly to (Kx/Lx)
other hand the domestic prices of n
ctly affected by trade; trade does ha
he following three ways.
goods are close substitutes to imp
s from nontraded goods to ch
the factors employed in production
return to them will fall.
goods are complementary to imp
ntraded goods will rise, increasing th
Capital-Labour ratio
SS
i
n
2
v
e
Y
X
(Ky/Ly)2
(Ky/Ly)1 (Kx/Lx)2 (Kx/Lx)1 (Px/Py)1
X
(Px/Py)2
(w/r)2
(w/r)1
Relative Price of
17
aded determine
ice of X (labor-
tal, w/r, up in a
elative price of
of good X is
t is at (Ky/Ly)1.
y)2, the ratio of
creased labour
tion of goods X
and (Ky/Ly)2,
ontraded goods
e a bearing on
orts, domestic
aper imports;
of nontraded
orts, domestic
eir prices. Thus
18
the factors employed in the production of nontraded goods will
gain as return to them will rise.
• If nontraded goods are neither substitutes nor complementary
goods to the imports, then there is no impact of trade on prices of
nontraded goods, thus no change in the return to their factors of
production is likely to take place.
Precisely, in light of the above described theoretical milieu, trade
liberalization would benefit the sectors that use the country’s most
abundant factor (labor in the case of Pakistan) intensively in their
production and harm those sectors that use the country’s scarcest factor
(capital in the case of Pakistan) intensively in their production or the
sectors that produce those nontraded goods that are close substitutes to any
of the imports.
Traditional Ricardian Theory suggested that only labor, as a single factor
of production, is needed to produce goods and services. Due to variation in
the technology across nations, the labor productivity is different among
different nations. It was this difference in the technology that initiated
advantages in producing specific goods and trading. Some goods or
industries are capital intensive if more capital per unit of labor relative to
other goods is used in their production or they have a higher capital-labor
ratio than other goods or industries in the country. Similarly, there are
goods or industries that are labor-intensive if more labor per unit of capital
relative to other goods is used in their production or these industries or
goods have a lower capital-labor ratio than other goods or industries.
19
2.3.2 Implications of the Model
Adjustments in national trade policy bring about changes in the prices of
goods (traded and nontraded) consumed domestically and in the labour
incomes11 of the workers. The impact of trade reforms comes from import
and export sectors.
In restricted trade regimes, prices of imported goods are kept higher than
the world price by imposing tariff and non-tariff barriers to trade. Liberal
trade policy may tend to increase the economic activity in the liberalized
sector as competition wipes out distortions from the market, paving the
way for efficient allocation of resources, trade liberalization on the other
hand may bring about losses for the local producers as they may lose their
share in the domestic market. Improved functioning of local markets due
to competition and less government intervention which helps in generating
new livelihood opportunities, reduces price and supply variability of
commodities, and eliminates market distortions
(monopolies/oligopolies/price administering, etc). Further, the imported
machinery, raw material, and advanced know-how may lead to enhanced
efficiency in the domestic production sector and increase the rewards to
the factors of production. Or, firms producing under the earlier protection
may lose hold of their previous market and embark on layoffs and
downward adjustments in the returns to their factors owing to the
competition. Similarly, the removal of trade barriers promotes export of
local products to the world market. The rise in exports cuts the existing
supply of a good in the local market tending to raise the domestic price.
Rise in the price of a good improves profit prospects for the business and
helps its expansion. The expansion of the producing unit results in higher
rewards for the workers of the unit. In the case of a developing country,
11 Wages and working hours
20
the returns to labor (wage) used intensively in the production sector would
rise, and returns to capital (rent) tend to fall.
In general, if liberalized trade policy affects the supply and thus reduces
the domestic prices of goods that are part of the consumption basket of the
poor in the country, the policy seems to benefit household welfare.
However, the fall in domestic prices of goods would in some ways affect
the labour incomes of the workers. Therefore it would be quiet unrealistic
to assess the impact of liberalized trade on poor household welfare just
from the information about change in the domestic prices. The realistic
assessment of the impact of trade on household welfare would consider
the cumulative impact of free trade on domestic prices and labour incomes
of workers.
2.4 Evidence from the Literature
For better organization and easy comprehension of the historical evidence
on the issue, the existing literature regarding trade effects on poverty via
factor and goods prices and household incomes in developing economies
is divided in two segments. The first segment consists of various studies
devoted to the impact of liberalized trade policies on wages of unskilled
workers in Latin American countries and the second studies the same, but
for Asian countries. Open trade experiences in the two regions have
encountered a situation of conflict of evidence. Latin American countries
suffered increased skilled-unskilled wage inequality and a rise in poverty,
and East Asian countries enjoyed a noticeable drop in wage differentials
leading to a reduction in poverty. The following sections present the
experience of Latin American and Asian countries’ trade policies.
21
2.4.1 Evidence from Latin America
Demonstrably, Latin American countries’ case is counted a failure of open
trade policy in light of the theoretical implications of the H-O model. The
reckoning stems from the fact that the skill premiums rose and inequality
and poverty worsened in these countries with the implementation of open
trade policies. During the late 1970s and 1990s many Latin American
countries (namely Costa Rica, Mexico, Chile, Colombia, Argentina, and
Uruguay) implemented an open trade policy by lowering tariffs and easing
quantitative restrictions on imports. Consequently, the skill differentials in
wages (identified at the levels of education) widened contrary to the
conventional wisdom of the H-O theorem12. The widening occurred from
the mid-1970s to the early 1980s in Argentina and Chile and between the
mid-1980s and the mid-1990s in Colombia, Costa Rica, and Uruguay. In
all cases, the relative number of skilled workers was rising, and thus the
dominant influence of the change in wages was a rise in skilled labor
demand. Time series calculations made by Wood, A. (1997) confirmed
that the relative demand for skilled workers rose during the liberalization
episodes in these countries. Skill differential in wages widened after the
mid-1980s in parallel with radical liberalization of the trade regime in
Mexico. Other studies have also explored the issue and confirmed the
presence of an association between wage inequality and open trade
policies in Latin American countries. Hanson and Harrison (1999)
estimated a trade-wage inequality link for Mexico and found evidence that
the skill-based wage differential was a consequence of removal of tariff
restrictions from the sectors that were relatively intensive in the use of
unskilled labor. The unskilled labor abundant sectors had shrunk and the
relative demand for skilled labor had shown a rising trend. They found
little variation in employment levels but a significant rise in skilled 12 For survey of literature on Latin American experience of trade liberalization and causes of widening gap between skilled and unskilled premiums see Wood, A. (1997), Chaudhuri, S. and Ghosh, A. (2001) and Robbins, D. J. and Gindling, T. H. (1999).
22
workers’ relative wages in Mexico. On the other hand they found no
correlation between the intensity of skilled labor and changes in relative
product prices, as suggested by the S-S model.
Robbins (1994, 1994a, 1996) and Feenstra and Hanson (1997) concluded
their analytical studies with similar results. Feenstra and Hanson (1997)
argued that the growth in foreign direct investment, which is positively
correlated with the relative demand for skilled labor, led to the higher skill
premiums in Mexico. Robbins (1994) found evidence of wage dispersion
in Chile between 1975 and 1990. He found a positive link between wage
differentials and the rise in the demand for skilled labor in Chile. Beyer et
al. (1999), using a time series approach, also found a long-term correlation
between openness and wage inequality in Chile.
Another study, [Chaudhuri, S. and Ghosh, A. (2001)] collected the
literature on the Latin American experience with open trade policies, and
the authors concluded their analysis with the important statement:
“removal of tariff restrictions from unskilled labour intensive sectors left
them unprotected which were highly protected previously and rise in
capital receptive foreign direct investment are the liable elements for
increase in the skill premium and wage differential as a logical outcome of
trade reforms.”
Additionally, Ianchovichina et al. (2001) used two-step procedures to
study Mexico’s potential unilateral tariff liberalization impact on Mexican
households. In first step they used Global Trade Analysis Project (GTAP)
model as the new price generator and in second step they applied the price
changes to Mexican households’ welfare. They concluded with a positive
effect of trade reforms on all income groups.
23
Without falling into a methodological controversy of evidence and
challenging the individual research work thereby, one may pose a serious
question here: Is it the liberal trade policy that intensified the skilled-
unskilled wage differentials or is something important missing from
consideration in the studies, for example the time of opening of the
economies and other methodological factors, affecting the findings. The
issue of conflict in evidence will be dealt with in the section on discussion
of empirical evidence. It is, though, difficult to develop a generalized point
of view on the impact of trade in favoring the unskilled production factors
in developing countries, but it is equally hard to ignore a consistent,
significant, and important factor of open policies resulting in a reduced
skill-unskilled wage gap in East Asian countries. The following section is
devoted for the East Asian experience with open trade policies.
2.4.2 Evidence from Asia
The evidence from so-called East Asian tigers (Hong Kong, the Republic
of Korea, Singapore, and Taiwan) on a trade-poverty link supports the
standard view of the H-O trade model, that the acceptance of more open
trade policies in developing countries with large numbers of unskilled
workers leads to increased demand for workers with a low level of skill
and education relative to the demand for highly skilled workers. The wage
gap between skilled and unskilled workers in South Korea and Taiwan
narrowed during the 1960s and in Singapore during the 1970s, and from
1973 to 1989 in Malaysia (Robins 1994a) after adoption of more open
trade policies. These countries had adopted open trade policies during the
1960s and 1970s and so gained the status of “early globalizers”. China,
though, joined the globalizers’ club during the early 1970s. Its rank
steadily rose from 30th largest trading country in 1977 to 3rd largest
importer (after EU and US) and 2nd largest exporter (after EU) in 2010
[WTO statistics 2010]. The most common aspect of these East Asian
24
newly industrializing countries is that they have been in the direction of
liberalization all along. There have been continuous unilateral trade policy
reforms in these countries away from high levels of protection previously.
However, Hong Kong and Singapore can be an exception in the group of
Asian emerging economies because they have been free port economies,
practicing zero import or export restrictions since the 1950s. These two
countries share a striking similarity with other East Asian Tigers that they
are at relatively the same level of economic development as each other.
Further, two more studies Fields (1994) and Robbins and Zveglich
(1995a) can be good sources for demonstrating the experience of these
four East Asian countries from open trade policies and poverty reduction.
Fields (1994) found that labor market conditions improved in all four
economies during the 1980s at rates on par with the rates of their
aggregate economic growth and that they grew without any repressions on
their labor markets during the same period. The four-country average rate
of growth in real per capita GNP is reported as 87.875% during the 1980-
90 decade, with a 89.57% growth rate in the real earnings of workers in
different sectors13. These East Asian newly industrialized economies
experienced a reduction in wage inequality after openness, with a strong
export-orientation introduced in the 1960s and 1970s. This was therefore
consistent with standard trade theory, which predicts that trade
liberalization should benefit the locally abundant factor (Wood, 1995,
1997; Krueger, 1983, 1990).
13 Hong Kong: Growth in real GDP per capita (64.2%): Earnings in manufacturing (60.0%); Korea: Growth in real GDP per capita (121.8%): Earnings in manufacturing and mining (115.8%); Singapore: Growth in real GDP per capita (77.5%): Earnings in all industries (79.8%); Taiwan (China): Growth in real GDP per capita (88.0%): Earnings in manufacturing (102.7%). Source: For Hong Kong: Government of Hong Kong (various years); for Korea: unpublished country data; for Singapore: Government of Singapore (1990); for Taiwan (China): Government of China (1991b). Also cited in Fields (1994).
25
2.5 Discussion on Empirical Evidence
The key implication of the difference in timing of embracing open trade
policy stems from the fact that by the time the Latin American countries
adopted open trade policy, they had lost the comparative advantage of
countries being rich in unskilled labor. Entry in the global market of four
heavily populated Asian countries—namely Bangladesh, China,
Indonesia, and Pakistan—with large numbers of unskilled workers by the
mid-1980s altered the position of Latin American countries in receiving
the comparative advantages from international trade. Although the ratio of
skilled to unskilled workers in those countries (Latin American) was still
far below that of the developed world, it was still above the global
average. This changed the basic principle of comparative advantage for
Latin American countries from the production of goods of low skill
intensity to goods of intermediate skill intensity. Additionally East Asian
countries opened up in the 1960s had already accumulated enough
skills/capital to shift their comparative advantage too from low to
intermediate intensity skills goods. Thus the greater openness in Latin
American countries during the 1980s instigated contraction of the sectors
both of high skill intensity goods (by imports from developed countries)
and of low skill intensity goods (by imports from low income newly
globalizing countries). The net effect might have been in either direction,
but greater openness could only result in an ever wider gap between
skilled and unskilled workers’ wages. The above explanation is supported
by Kaplinsky (1993), who attributed the losses in labor-intensive Latin
American manufacturing sectors to competition from imports from low-
income Asian countries domestically and in third-party market (for
example in the US, an important destination for Latin American
exports)14.
14 The main inspiration for the implications here is drawn from Wood, A. (1997).
26
Further, two more plausible explanations of the conflict of evidence on
trade liberalization experiences in the two regions have been explored in
Wood, A. (1997). The first implication is related to the increased global
demand for skills. Citing Robbins and Zveglich (1995a), Wood quotes the
global skill demand as the “Skill Enhancing Trade”. However, this
explanation is not ironclad, as the opening countries were not completely
cut off from new technology, yet, it is most likely that the countries
accumulate skills and alter their technological demand with increased
openness. East Asian countries can be presented as a reasonable case study
in this respect.
The second implication is associated with the difference between East
Asian and Latin American natural-resource endowments. In East Asia,
during the 1960s the majority of exports were concentrated in
manufacturing, whereas in Latin America trade gains were emanating
mainly from primary and processed primary exports, with manufacturing
exports often shrinking, except in the parts of Mexico adjacent to the
United States. This was because Latin America is far better endowed with
natural resources than East Asia and consequently had a comparative
advantage in production of primary products. Therefore the claims of
failure of liberal trade policies in Latin American countries may be refuted
with the rationale that it was not the trade reforms that raised skill
premiums but rather the increased global demand for technology, entry of
many labor-abundant countries in the world trade sector during the 1970s
and 1980s, and richness of the Latin American region in resource
endowments.
27
3 Choice of Methodological Technique
This Chapter opens the discussion on the choice of methodological
technique for the analytical structure of the present study. The previous
literature on the trade effects on various economic parameters
contemplates diverse methodologies that differ in a number of significant
ways. These analytical studies vary across dimensions, with analysis
carried out for representative households or actual households, employing
dynamic or static analysis, using single- or multi-regional statistics, and
using partial or general equilibrium approaches. Of these possibilities, four
main categories are identified as the important techniques, based on the
principal methodology applied [Reimer (2002)]. Each technique inherently
has certain limitations and degrees of complexity along with certain
advantages when applied for a variety of research objectives. The choice
of a suitable technique in a research study depends upon the desired
outcome and adherence to certain intrinsic conditions and limitations the
techniques are subject to. These conditions can be related to availability of
required quality data, accessibility to computing resources such as
computers with specific programs/softwares, and expertise required when
more complex and larger models are employed. Further, it is also crucial
to undertake the analysis of intended objectives before choosing any
methodology, whether the target is to measure the aggregated welfare
impact of a policy shock on the whole economy or is restricted to
exploring the relationship between a certain policy shock and a particular
variable. Since the conditions, scope, circumstances, and targets of
research projects vary in goals, so is the case with research techniques.
Hence, the preference for one research technique over others remains an
important area for authors and researchers as far as their own objectives
and goals are concerned. The principal methodological techniques
explored are Computable General Equilibrium Modeling technique, Partial
28
Equilibrium Analysis and Micro-simulation Modeling technique. Cross
country Regression Analytical Methodology would be out of the scope of
the present study since the present study is about working with national
data from a single country, Pakistan.
Thus the core discussion in this section encircles the merits and demerits
of the above three research techniques/methodologies. The purpose is to
justify the selection of one out of the three techniques, allowing realization
of the present study’s goals and advantages, limitations and conditions
embedded in the use of all three techniques individually.
The following descriptions of Partial and Computable General
Equilibrium are based on the textbooks of Black, Fischer (1996), Mas-
Colell, A., Whinston, M., and Green, J. (1995), and Varian, H. R. (2003).
For a detailed literature survey and categorization of studies by use of the
main research technique, refer to Reimer (2002).
3.1 Computable General Equilibrium (CGE) Analysis
This modeling technique represents a powerful tool used for
distinguishing the multiple economic effects on an economy surfacing
from various economic and trade policies. The CGE model addresses the
workings of an economy in an integrated manner by considering the
complex inter-linkages and feedbacks between production sectors,
households, and institutions.15
The following paragraphs will present an overview and the working of
General Equilibrium theory and model from the perspective of historical
and pioneering contributions by L. Walras (1834-1910), V. Pareto (1848-
15 For a brief history of General Equilibrium Modeling technique and a survey of its main contributions and application of the technique see Borges, Antonio M. (1986).
29
1923), F. Y. Edgeworth (1845-1926), and I. Fisher (1867-1947) in giving
the CGE technique its present modern shape. The fundamentals of the
CGE theory and model are provided by General Equilibrium theory,
which was introduced by French-born mathematical economist L. Walras
(1834-1910), a prominent marginalist and professor at University of
Lausanne, Switzerland, in his book Elements of Pure Economics published
in 1874. According to his theory, in a market system the prices and
production of all goods are interrelated. A change in the price of one good
is likely to alter the prices of other goods in the society. For example, a
small change in the price of bread may change the wages of the workers in
the bakery. Owing to these links between individual economic agents
(markets and households) in the economy, the theoretical calculation of
equilibrium price of just one good requires an analysis that accounts for all
of the various goods that are available in an economy.
Because the theory studies the behavior of individual agents in an
economy toward any policy changes and is capable of analyzing issues at
a micro level, this is distinguished as part of theoretical microeconomics
using a bottom-up approach (from analyzing links between individual
economic agents at the bottom to the whole economy at an aggregate
level). This microeconomic foundation of CGE specification guarantees
the simultaneous interaction among micro, market, and macro levels of the
economy that can capture all horizontal, vertical, and forward-backward
links among all production sectors, factors of production, and households
in the economy. See detailed account of theory of General equilibrium in
Kuenne, R. E. (1963).
The prevalence of perfect competition in the market is the key assumption
of General Equilibrium theory. Each decision-making unit in the economy
operates independently, i.e., each firm acts as if it were trying to maximize
its profits and every household acts as if it were trying to maximize its
utility. Thus the theory, in the market economy, seeks to find such a
30
unique solution where each unit of output is produced and sold at its
lowest unit cost in the quantity demanded by each household, given that
all markets are cleared. Assumption of perfect competition in the market
further suggests that each economic agent is a price taker and that all
prices are flexible. More precisely, the theory seeks to explain production,
consumption, and prices in a whole economy by coordinating the choices
of all economic agents across all goods and factor markets. Since all
markets are interdependent, simultaneous solution to the system implies
that, as mentioned above, the price of any one good will be affected by a
change in the price of the other good. In addition, by virtue of production
and market theories, it is assumed that the system is homogenous of
degree zero in absolute prices: if the values of all price variables are
increased equi-proportionately, the values of the quantity variables will be
left unchanged. The main issue, thus, is the existence of equilibrium in all
sectors of the economy. That is, even though it could be demonstrated how
individual markets behaved, it would remain unknown how goods
interacted with each other to affect supplies and demands in multiple
markets in the absence of a simultaneous solution for all markets.
While working on his book Elements of Pure Economics (1874) Walras
presented his idea of equilibrium by conceiving the prevalence of
consistency in the equilibrium concept in terms of the number of equations
required for market clearing and the number of variables available to
obtain it: the prices [Carvajal, A. (2006)16]. To solve this problem, he
created a system of simultaneous market demand and supply equations.
Studies using Computable General Equilibrium modeling technique
account for commodity market and terms of trade, and factor market
effects by using disaggregated Social Accounting Matrix (SAM) as an
analytical base. Several studies have been conducted using CGE modeling
16 p1
31
technique to measure the trade effects on poverty and welfare in
developing countries. Some of these studies include Coxhead and Warr
(1995), who examined the impact of technical progress in agriculture on
changes in poverty and aggregate welfare in Philippines. Loefgren (1999)
analyzed the short run equilibrium effects of reduced protection in
agriculture sector using GE model and found that the reduced agricultural
protection would generate significant aggregate welfare gains. Cogneau
and Robilliard (2000) used general equilibrium framework to examine the
impact of various growth strategies on poverty and inequality prospects in
Madagascar. Sadoulet and De Janvrry (1992), have pursued a multimarket
approach for analyzing the impact of trade liberalization on the agriculture
sector in Africa using General Equilibrium approach. Harrison,
Rutherford, and Tarr (2001) explored the case of trade liberalization and
poverty in Turkey using CGE model with 40 households distinguished by
income levels and urban or rural locations. Evans (2001) worked to
investigate the impact of global trade policy reform on South Africa by
integrating the findings from GTAP and poverty case studies for Zambia.
He found that the unilateral trade reforms improved income but were
having strong bias towards metropolitan areas against poor rural sectors.
Limitations and Advantages
Traditionally, for three obvious reasons, the study of General Equilibrium
analysis has been emphasized: first, it studies the essential duality of
pricing and resource allocation; second, it represents the interdependence
of different parts of an economic system; and third, it provides a unifying
framework within which some major branches of economic theory such as
the theory of value, welfare economics, pure theory of international trade,
and the theory of economic growth can be shown as having a common
32
origin, since all have the common goal of determining the price of goods
and services and efficient allocation of resources [Simpson, D. (1975)17].
Though theoretical superiority of Computable General Equilibrium system
has remained unchallenged, nevertheless some studies have argued that
analytically it is not a useful exercise and found it limited to merely the
description of numbers and data without giving a concrete basis for policy
making. Borges, A. M. (1986) concludes that CGE models happen to be
significantly large, comprise substantial parameters, and often embody
complex structures. Parameters incorporated into the model are not
estimated econometrically; rather, they are estimated independently out of
the model and are then calibrated to a single data point, which is chosen to
represent a situation close to general equilibrium [Borges, A. M. (1986)18].
Thus the exercise of parameters being isolated from the main model leaves
the results of the model not to forecast the reality but rather only to
indicate long-term tendencies around which the economy will fluctuate.
Due to that fact, results from CGE modeling can neither be useful for
replication of the evolution of the economy in the past as a means of
checking their validity nor can be applied for future policy making. This
feature of CGE modeling defeats the inherent purpose of the research, i.e.,
performing efficient future policy making based on concrete estimations,
results, and evaluating the previous policies by using trade models.
Its strengths include its coherent microeconomic theoretical foundation,
internal consistency, suitability for policy issues involving substantial
changes in variables’ absolute and relative terms, and its concern to
measuring welfare loss or gain of the whole economy. But these are
merely theoretical advantages, since we could never include every aspect
of the world economy in a mathematical model, nor could we quantify
every step of certain policy implementation precisely in any computer
17 p 9 18 p 19
33
simulation model. Therefore, performing CGE exercises without
econometrically estimating the coefficients and parameters is nothing
more than scientifically pretending to cover all of the linkages and
feedbacks from the whole economy in the analysis, while the reality
happens to be far from it.
3.2 Partial Equilibrium Analysis
Partial Equilibrium Analysis is a way of obtaining an estimate of the
impact of a change in the economy that does not require the complete
solution of the General Equilibrium system [Whalley (1974]. It is another
view of measuring and establishing the link between variables (for
example, trade and poverty) in the economy. A Partial Equilibrium view
in the context of trade is considered a part of the General Equilibrium
analysis, where the clearance of the market of some specific goods is
obtained independently from prices and quantities demanded and supplied
of other goods' markets. Unlike CGE models with an aggregate behavior,
Partial Equilibrium models of trade do not give an aggregate view of the
welfare of an economy; rather, they allow researchers to focus on how the
gains and losses from a shift to free trade are shared across specific
individuals/households and markets in a more detailed and reliable way.
The argument can also be put this way: It is a way of obtaining an estimate
of the impact of a change in the economy without requiring the
simultaneous solution of the whole economic system. The impact of real-
world policy options on any specific sector is investigated, keeping other
things constant under ceteris paribus assumptions. Quoting the trade-
poverty link here, one can assert that the investigation of the impact of
trade policy on a specific sector is established through disseminating the
straightforward way of measuring welfare effects of international trade
34
through estimating changes in consumer surplus, so that consumer welfare
can be measured.
This type of analysis either ignores effects of the policy in other industries
in the economy or assumes that the sector in question is very small and
therefore has little, if any, impact on other sectors of the economy.
Whaley (1974) classified Partial Equilibrium Analysis into simple and
extended versions. According to him, under Simple Partial Equilibrium
Analysis, all prices and quantities except of the commodity under
consideration are treated as constant and non-variant with time.
Additionally, linearization assumptions are employed as local
approximations to ease the problem of computing new estimates after
policy implementation. This is also known as the log linear version of the
system, due to its linearity assumption. However, he reckons it an
Extended Partial Equilibrium Analysis when the linearization assumptions
are relaxed and the impact of the change of a single price (when allowed
to vary) upon the value of the demand for other goods via changes in the
value of endowments is incorporated inside the system environment. This
nonlinear version of the system can be solved through linear
approximation methods [Borges, A. M. (1986)].
Partial Equilibrium technique is deemed useful in studies that focus on
specific and straightforward relationships between variables with strong
precision for future policy making. Use of this technique can help
overcome various research-related issues, such as availability of large
accurate and reliable data.
There are ample studies using Partial Equilibrium modeling approaches.
By using household expenditure data, these studies generally emphasize
commodity markets and their role in determining poverty impacts as a
measure of poverty across time. The studies reviewed here using this
approach are Appleton (2001), Fofack, Célestin, and Tuluy (2001), Deaton
35
(1989), Dercon (2001), Ravallion (2004), Ravallion and Van de Walle
(1991), Levy and van W. (1992), McCuloch and Calandrino (2001), Case
(1999), Levinsohn, Berry, and Friedman (1999) and Minot and Blauch
(2002).
Limitations and Advantages
Some macro-level studies using Partial Equilibrium technique divulge the
inability of this technique to precisely measure the terms of trade effects
for each region and assess their income distributional consequences across
regions and socioeconomic groups within regions [Harding, Ann
(2007)19]. Partial Equilibrium technique has limited ability to assess
second-order effects (inter-industry effects and macroeconomic
adjustments that often appear to be significant) of a policy. However,
Partial Equilibrium technique is preferred when the sector under study is
only part of a whole, so the generated effect claims are exclusively and
precisely for that specific part of the whole. Further, the technique is
efficient when the shock from a policy change to be measured on a sector
has limited backward and forward linkages with other sectors of the
economy.
It focuses on only part of the economy at a time, overlooking interaction
between various markets in order to overcome the complexities (arising
from limited availability of accurate data and use of complicated computer
programs, for instance) and ensuring simplicity, straightforwardness, and
transparency in analysis owing to its reliance on few key parameters. The
above arguments are considered as the benefits of this system, rendering it
preferable to Computable General Equilibrium technique. The partial
19 p5
36
system is more reliable and authentic vis-à-vis its usefulness in future
policy-making and evaluation of trends in past policies.
The present study uses fairly long time series data of 36 years for demand
estimation and 14 years data on tariff to investigate and estimate the links
between prices, labour incomes, and poverty to analyze past trends and
perform future forecasting. Since all relationships between various
parameters are econometrically estimated therefore, results derived under
Partial Equilibrium analysis are trusted for future forecasting and policy
making, since they are more reliable and straightforward than those
derived from Computable General Equilibrium model.
3.3 Micro Macro (Simulation) Models
The need for spatially disaggregated data gave birth to this technique.
Spatially disaggregated data can be extremely useful for regional and
social policy analysis, as it presents efficient representation of individuals.
It is a technique used to model complex real life events by simulating the
impact of policy change (characteristics and behavior) on individual units
of the whole system wherein the changes occur. It has generally been
accepted as a valuable policy tool used to analyze the detailed
distributional and aggregate effects of both existing and proposed policies
at a micro level, where individual households are taken into account to
capture individual heterogeneity. Pioneering work on micro simulation
was conducted by Orcutt (1957)20.
In the modern world, researchers and policy makers attempt to achieve
multiple social policy objectives such as income redistribution, ensuring
access to health and education, and a reasonable standard of living for
most of the citizens. Most researchers prefer solo microsimulation models
20 Also cited in Merz, J. (1995)
37
to assess distributional effects of public policies with no possibility to
analyze the efficiency impact of the policy. Others prefer macro models to
measure the efficiency of the policies while ignoring the distributional
effects. Given the emphasis on changes in income distribution,
microsimulation models are often used to investigate the impacts on social
equity of fiscal and demographic changes (and their interactions).
Modeling of the distribution of traffic flows, for example over a street
network, is another increasingly important use of the approach21.
For the purpose of distributional analysis of micro data, micro simulation
models are integrated with macro models. These models/links can serve
the dual purpose of computing efficiency impacts and conducting
distribution analysis using micro data. Two approaches have been
identified in incorporating micro data into a macro model: Integrated
Microsimulation models and Micro-Macro models. Integrated
Microsimulation models attempt to incorporate individual household
information generally found in income-and-expenditure-based household
surveys into macro frameworks. Data for these models comes in the form
of disaggregated SAM account. The labor- and wage-based income
generation of different households is categorized according to various
occupations in industrial sectors, profit income, government transfers, rest
of the world transfers and other income. A Micro-Macro approach follows
mainly sequential linking of a model based on micro-level data with a
model based primarily on macro-level data [Reimer (2002)]. The key
studies that use this technique are Ianchovichina, Nicita, and Soloaga
(2001), Cockburn, J. (2001) and Robilliard, Bourguignon, and Robinson
(2001). The models are then linked by modifying selected parameters of
the Microsimulation model according to certain variables generated by the
macro model.
21The International Microsimulation Association descriptions
38
Microsimulation model is used to replicate microeconomic features of a
labor market as well as household consumption and income behavior from
household data sets, while a macro model generates values for macro
variables such as total employment, prices of commodities, and wages,
etc. Finally the Microsimulation model is solved in such a way that results
are consistent with the aggregated variables generated by the macro
model. This is different from the previous microsimulation approach in
several ways. In the earlier method, the process of disaggregating
individual households from the household sector captured micro-level
issues, whereas in this approach a separate model is developed
encompassing other socio-economic and demographic features to capture
the interactions explicitly.
Limitations and Advantages
Comprehensive data linkage, spatial flexibility, and the ability to update
existing data and forecast for the future are some of the advantages of
Micro-Macro modeling technique. However, its disadvantages, such as
difficulties in calibrating the model and validating the model outputs
sometimes even overshadow its advantages.
3.4 Choice of an appropriate Modeling Technique
The goal of the study is to provide reliable econometrically estimated
results from a long time series of 36 years data for the demand estimation
and 14 year data for measuring the price effects and change in labour
income in agriculture and manufacturing, which can be used for future
forecasting and analyzing past trends regarding the link between trade and
39
poverty in Pakistan. The nature of the study in hand requires
accomplishment of a two-fold goal. First, establishment of the link
between reduced import duties (trade openness) on poverty in the country
via tracing changes in labour income in the selected sectors and prices of
selected traded and nontraded goods consumed by the poorest households
in Pakistan. Second, generation of econometrically tested welfare
estimates that can be used for future policy making as well as have the
capability to explicate past trends. Use of Partial Equilibrium approach
(with support of general equilibrium framework) in the study seemingly
discharges the two-fold goal in a more efficient way than the CGE and
Micro-Macro techniques. The following points support the preference of
Partial Equilibrium technique over other techniques.
1. The present study is about analyzing the long time series data of 36
years to estimate demands and 14 years time series to estimate the
welfare measures and change in the labour income in agriculture
and manufacturing, and CGE modeling technique does not provide
the empirical validation since there is no measure of the degree to
which a model can fit the data or can track historical facts. Owing
to the unrealistic assumption of prevalence of equilibrium in all
markets, the CGE technique is inappropriate to be used for
forecasting the reality; rather, it can only indicate the long-term
tendencies around which the economy will fluctuate.
2. Partial Equilibrium technique is preferred for its simplicity since it
does not call for developing complex structures based on
accounting and input-output matrices that are needed in General
Equilibrium technique to obtain the required results. Since the
research is devoted to the enquiry of links between specific
variables in a part, rather than the whole, i.e., the link between
trade policy changes and changes in labour incomes and prices.
Partial Equilibrium modeling technique, thus, suits the study most
40
as it produces estimated results usable for future policy forecasting
and evaluating past trends.
3. Again, the purpose of the study is not to measure the terms of trade
effects for each region and assessment of income distributional
consequences across regions and socioeconomic groups within
regions, where Partial Equilibrium has some limitations and
General Equilibrium technique has an edge over Partial
Equilibrium technique. Yet, the limitations of Partial Equilibrium
technique in this specific realm would not hamper the realization
of the goals of the present study.
4. The estimation of second order effects of trade policy across the
economy is out of the scope of the study. So use of Partial
Equilibrium technique would still have no bar on attainment of the
objectives of the study. Though General Equilibrium modeling
technique is more suitable while considering second order effects,
the General Equilibrium results are not econometrically estimated
and so not reliable for future policy making and forecasting.
Therefore, Partial Equilibrium modeling technique seems more reasonable
and reliable since the feature of General Equilibrium technique to carry
parameters into the model that are not estimated econometrically, rather
are estimated independently out of the model and are then calibrated to a
single data point, which is chosen to represent a General Equilibrium
situation, limits its capability to be used for future policy making and
analyzing past trends. Further, the use of a Micro-Macro technique would
be more complex and cumbersome than necessary for the present study
[Borges, A M. (1986)].
41
4 Trade Liberalization, Prices of Traded and Nontraded Goods, Households’ Labour Income, Welfare, and Poverty
The idea here is to investigate the distributional effects of trade reforms
and a selective protectionist trade policy in some commodity groups on
the welfare of the poorest households in Pakistan via changes in domestic
prices and labour income using household and labour force data in partial
equilibrium setting22. The approach is based on the specifications of Porto
(2003, 2006), however some of its methodological inaccuracies are also
pointed out and corrected in the present study. Porto (2003) implicitly
assumed that there are no quantity effects of a trade liberalization induced
price change. That would imply that the households’ demand curves for all
selected goods are vertically sloped which is against the conventional
demand theory. Secondly, Porto has estimated the domestic prices on the
international prices and the tariff rates treating implicitly the tariff rates as
a variable determined in the system whereas on the contrary the tariff rates
are fixed by the government so are exogenously given like the
international prices. In present study these failings have been corrected by
first considering the quantity effects by allowing the quantities demanded
of the selected goods to change with a change in the domestic prices.
Second, the domestic prices are not estimated instead they are calculated
by adding the tariff per ton to the international prices. The following
analysis begins when a change in the trade policy, treated as an
exogenously determined variable, brings about a change in domestic
prices of traded goods. This change in domestic prices further entails a
multitude of other impacts in the small open economy, including a change
in the domestic prices of nontraded goods and the labour income. The
22 For review of the studies examining the trade liberalization effects using household data in partial equilibrium see Attanasio, O., et al., (2004) and Deaton, A. (1989).
42
variations in domestic prices and the adjustment in labour income23 lead to
a change in households’ welfare.
The present Chapter is organized in two parts. The first part is about the
modeling of some basic interrelations between trade policy, domestic
prices and factor prices. The second part of the Chapter includes the links
between trade policy, households’ demands and welfare effects. The
welfare effects of a change in the trade policy are captured by measuring
the change in Marshallian Household Consumer Surplus. The Marshallian
Household Consumer Surplus is then corrected by deducing the income
effect from the total price effect using the slutsky equation to measure the
Hicksian Compensating Variation. Further, the change in labour income of
agriculture and manufacturing workers resulting from the change in the
trade policy is measured. The change in the poorest households’ welfare is
measured using Marshallian Consumer Surplus approach.
4.1 Interrelations between International Trade, Domestic Prices, and
Factor Prices
4.1.1 Domestic Prices of Traded Goods
Since the price variations accounted for are trade and tariff driven, it
would be interesting to know how domestic prices of traded and nontraded
goods are determined in the local market when the tariff rate changes.
Pakistan, being an economically small developing country, plays the role
of a price taker in the global trade sector. Thus in analogy with other small
open economies, the determination of domestic prices of traded goods in
Pakistan would look as follows:
23 Wage rate times the working hours
43
i wi iP=P (1+t ) (1)
Here Pi and Pwi are the domestic and world prices of the traded goods i
respectively, and ti is the rate of tariff applied on traded goods. If the
international price is exogenously determined, then the change in the local
price would be established by the given change in the rate of tariff (which
is also exogenously determined as it is fixed by the government). This is
shown in the following set of equations.
)t(1P)t(1PPPdPi i1wii2wii1i2 +−+=−=
or
)(tPPPPdP i1i2wii1i2i −=−= (2)
or
iwii dtPdP = (3)
It can be inferred from equation (3) that given exogenous world price, the
absolute change in the domestic price depends upon the international price
times the tariff change. Taking log on both sides in the equation (3) we
will have;
)t(1dlnPdlnP iwi += (4)
For simplicity reasons, here we would allow the relaxation of two strong
assumptions. Firstly, there are unified products and one tariff line for
imports of the same product for all countries. In this way, we are indeed
relaxing the Armington assumption [Lloyd, J. P. et al. (2006)] of
differentiated products with respect to their various points of origin or
production (countries). Secondly, it is further assumed that the goods have
similar prices throughout the whole country. Though, in developing
countries, this assumption may not hold in its entirety for a variety of
reasons such as irregular market structures, information unevenness, etc.
44
Nevertheless, in the case of Pakistan, owing to sea access and a relative
good communication and transportation infrastructure, as well as
developed markets in urban sectors, equation (4) can be a reliable exercise
to determine the absolute price changes caused by a change in tariff.
4.1.2 Domestic Prices of Nontraded Goods
As we already have learnt from the S-S findings, the prices of traded
goods are dependent upon factor price (wage)24:
)(wfP iii = (5)
This means that if the factor prices can be derived from the prices of
traded goods, they in turn uniquely determine the prices of nontraded
goods. In this case, prices of nontraded goods are independent of demand
conditions or factor supply. They only depend on technology and the cost
of input factors. Thus, in general equilibrium, an aggregate relationship
between prices of traded and nontraded goods can be established in the
form of:
)f(PP TNT =
Here PNT are prices of nontraded goods and PT are prices of traded goods.
The above relationship between prices of traded and nontraded goods can
be specifically expressed in following way:
T
NTTNT lnP
lnP.dlnPdlnP∂∂
= (6)
The equation 6 captures the change in the domestic prices of nontraded
goods by the multiplying the elasticity of prices of nontraded goods with 24 Also proved in section 4.1.3, see equation 19
45
respect to traded goods T
NT
lnPlnP∂∂
with the percentage change in the prices of
traded goods TdlnP .
4.1.3 Households’ Labour Income
At this point we are ready to establish a link between trade reform-induced
price change and the change in households’ labour income via the impact
of a price change on wages and working hours. To begin with, it is shown
that households’ labour income is determined from the information on the
prevailing working hours and the wage of the labor. It can be written as:
wLYi = (7)
Here w stands for wage per hour, L stands for labor hours, and Yi is the
total labour income of household i. This section mainly follows the
theoretical foundations of International Trade Theory provided in
advanced textbooks of Dixit and Norman (1980) and Woodland (1982).
The discussion on the impact of changes in the prices of traded goods on
wage with Cobb Douglas Production Function is opened here. Cobb-
Douglas Production Function is used in the whole study, having constant
returns to scale where the market is dominated by a large number of
buyers and sellers.
βαKLQ = (8)
As per the assumption of the constant returns to scale, the Cobb-Douglas
Production Function as shown in equation 8 assumes: α + β=1. To find out
the average cost we would begin with the total cost function, which is
given as:
rKwLTC += (9)
46
Here TC stands for Total Cost, w stands for wage, r stands for capital rent
and K stands for capital.
The assumption of constant returns to scale further implies two
propositions:
Increasing the two input factors (labor and capital) by λ would result in
output supplemented by the same amount, i.e., λ
λQQλKLλλKλLλλKλL βαβαβαββααβα ==== + (10)
Similarly the total cost would rise by the same amount:
T C = w λ L + rλ K = λ (w L + rK )= λ T C (11)
The average cost function can be derived from the production function (8),
retaining the minimal cost combination. The minimal cost combination
requires equality between the Marginal Rate of Technical Substitution
(MRTS) and the ratio of factor prices (wage and capital rent) wMRTS=r
,
where MRTS is the amount by which the quantity of one input is to be
reduced, when one extra unit of another factor is used, so that the output
remains constant. According to the firm theory, the MRTS is the ratio of
values of the marginal productivities of two factor inputs (capital and
labor). Thus we have:
LQ
MPL ∂∂
=
Marginal Labor Productivity
KQ
MPK ∂∂
=
Marginal Capital Productivity
rw
==
∂∂∂∂
−
−
α1β
β1α
LβKKαL
KQLQ
(12)
Solving equations for labor (L) and capital (K), we have:
47
= +
βα
rw
QL α11
(13)
= +
βα
wr
QK β11
(14)
After inserting results from (13) and (14), we can derive Total Cost
function under the conditions of perfect competition.
+
=
αβ
βα
βα
βα
rwQTC (15)
The Average Cost function, thus, is as follows:
+
=
αβ
βα
βα
βα
rwQ
TCAC (16)
After close observation of the relationship in (16), one can find that the
average cost is a function of the factor prices and a constant
+
αβ
βα
βα
made up of production elasticities of labor and capital.
Further, it should also be noted that the average cost does not depend upon
the level of output Q, since in perfect competition there is no profit margin
and the price is equal to average cost. Therefore:
(K)rwP βα= (17)
Or transformed to linearity and ignoring the constant (K), we have;
βlnrαlnwlnP += (18)
The relation between factor and goods’ prices expressed in (18) can also
be put in more general terms:
48
r)(w,fP ii = (19)
Equation (19) confirms that commodity prices Pi are the function of factor
prices w and r. Since our main focus here is to determine the link between
factor price and commodity price in a developing country setting with a
labor intensive production sector, the role of capital is ignored.
In equation (19), Pi is the vector of commodity prices, fi is the average
cost function, (since in perfect competition there is no profit margin and
price is equal to average cost) and w and r are the vectors of factor prices
(wage and capital rent). From this model, it is attempted to show that the
prices of goods are determined from the set of wages. Nevertheless, the
main objective is to know: Can wage be uniquely determined from
information on commodity prices, i.e., )( pfw j = ? Here rises the
question of invertibility. To resolve the issue of invertibility (univalence),
the 2x2x2 H-O model is referred. From 2x2x2 H-O model, one knows that
the commodity price vector has the ability to determine the factor price
vector as long as there is no issue of Factor Intensity Reversal25. This, in
other words, means that factor prices do not depend upon factor
endowments and are also not affected by any variability in it. The
relationship between commodity prices and factor prices in this specific
fashion has been called as Factor Price Insensitivity26 in the literature. This
can simply be illustrated by mapping the average cost functions in a factor
price diagram as shown in the following two graphs of Figure 5.
25 Factor Intensity Reversals is a property of the technologies for two industries such that the ordering of relative factor intensities is different at different prices. 26 Leamer (1995) has emphasized that a sufficiently diversified small open economy has a national labor demand that is infinitely elastic. Also quoted in Slaughter (2001).
Fig. 5: Factor-Price diagram with unit Reversals and (right) with Factor Inten
The curves in Figure 5 in graph
are downward sloping and conv
negative slope:
βα1
α
wPr =
Note that equation (20) is thβαrwP =
Both graphs in Figure 5 feature
rent and wages) where P equals
in Figure 5 demonstrates the un
curves are intersecting each oth
any unique solution for (w, r) in
in graph II intersect twice (A a
P1=C1 (w,r) and P2=C2 (w,r) un
right depicts the presence of the
B show that industry one is mor
r
cost curves (left) with no Factor Intenssity Reversals
I and II are the average cost c
ex to origin, the r(w) is a funct
e inverse of the original equa
the combination of factor pric
the unit cost for each industry. T
ique solution for (w*, r*) as bot
er at a single point A. Whereas
graph II of Figure 5. The unit c
nd B), producing (w1, r1) and
it cost curves. Therefore the gr
factor intensity reversal. The po
e capital intensive at point A tha
w
w
II
I
r2
r r1
P2=C2(w,r)
P1=C1(w,r)
P2=C2(w,r)
P1=C1(w,r)
B
A
w1
w2 w*
A
r*
r
49
ity
urves that
ion with a
(20)
tion (17);
es (capital
he graph I
h unit cost
there is no
ost curves
(w2, r2) at
aph on the
ints A and
n industry
50
two and more labor intensive at point B27. This means that the ordering of
the relative factor intensities is different at different prices. The
determination of the unique solution (for w and r) depends upon the unit
cost curves intersecting each other only once. This is true when the two
unit cost curves portray the same curvature (same value for the
substitution elasticity of production factors). If the substitution elasticity
between two industries differs, hence the curvature of the unit cost curves
also differs, and the unit costs curves necessarily intersect twice (or none).
If both industries exhibit Cobb-Douglas type production functions, they
intersect only once, since both have the substitution elasticity of 1. In
absence of factor intensity reversal, the change in the commodity price
will cause a change in the factor price. With a rise in the commodity price,
the real return to the factor used intensively in its production will rise,
while the return to the other factor will fall. This reflects the core of the
famous S-S theorem.
Further, the above example of Cobb-Douglas Production Function can be
referred back to in the following way. Taking the total differential of price
equation (17) and dividing by P would provide the percentage change in
goods and factor prices.
βα
1βα
βα
β1α
rwdrrβw
rwdwrαw
PdP −−
+=
or
rdr
βw
dwα
PdP
+= (21)
To grasp the connotation more clearly, the equation (21) can be broken up
into three parts: P
dPis the percentage change in the prices of commodities,
27 See Feenstra (2004) for detailed explanation of reversals of factor intensities between two industries.
51
wdw
is the percentage change in the wages of workers, and r
dr is the
percentage change in the rent of the capital. Parameters α and β provide
the weights of wage and capital rent (of percentage change in wage and
percentage change in capital rent respectively) that the wage and capital
rent has on the price change of a good. In other words, they define the cost
shares that each factor has in production.
Denoting the percentage change by hat ^, the equation can be rewritten for
the two-good two-factor case as follows:
^
1
^
1
^
1 rβwαP += (22)
^
2
^
2
^
2 rβwαP += (23)
After solving the above two equations for ^r and
^w and arranging them in
matrix form, we can receive it as follows:
−
−=
⇒
=
^
2
^
1
12
12
^
^
^
^
22
11
^
2
^
1
P
P
αα
ββ
D1
r
w
r
w
βα
βα
P
P (24)
Where D denotes the determinant of the matrix on the left, which is
defined as:
212112212121 ββαα)α(1α)α(1ααββαD −=−=−−−=−= (25)
Equation (25) is true after referring to the assumption of Constant Returns
to Scale, where 1βα ii =+
To facilitate the solution of the matrix for wages (w) and capital rent (r),
the determinant has been taken as inverse of the matrix.
To analyze the goods’ price changes, we assume first that the industry I is
more labor intensive and secondly that the price of good 1 rises after trade
52
reforms. As per first assumption we have the value of
0ββααD 1221 >−=−= and secondly, 0PP^
2
^
1 >− . By using the above
information, we can solve for the changes in factor prices. Multiplication
of column vector of percentage change in prices of two goods
^
2
^
1
P
Pwith
the inverse matrix
−
−
12
12
αα
ββ will solve the matrix for impact on wages
(percentage change) and capital rent (percentage change). Consequently,
the labor wage and capital rent equations would appear as follows:
Wage equation:
−=^
21
^
12
^
PβPβD1
w (26)
Without affecting the uniqueness of equation (26) and the results thereby,
we have arbitrarily adjusted the parameters (α, β) and price vector (^ ^
1 2P , P )
in (26a) and (26b) for wages and (27a) and (27b) for the capital rent:
−+−= )PP(β)β(βPD1
w^
2
^
1112
^
1
^
(26a)
or
( )( ) ( )
>−
−+
−−
=^
1
^
2
^
112
1^
112
12^
P)PP(ββ
βP
ββββ
w (26b)
Capital rent equation
−=^
12
^
21
^
PαPαD1
r (27)
−−−=^
2
^
1221
^
2
^
PP(α)α(αPD1
r (27a)
53
or
( )( ) ( )
<−
−−
−−
=^
2
^
2
^
121
2^
221
21^
P)PP(αα
αP
αααα
r (27b)
The above set of equations [(26a), (26b), (27a) and (27b)] validates the S-
S theorem: Wages in industry I (recall that industry one is using labor
more intensively than capital) rose by even more than the rise in the price
of good 1. On the other hand, the change in capital rent is even lower than
the change in the price of good 2. Further, it can be observed from the
above results that the percentage change in wage is higher than the
percentage change in the price of good 1. It can be shown in terms of real
wage and rent, too: ↑↑21 P
w,
Pw
. Real wage in terms of price of both goods
has increased, and capital rent in terms of prices of both goods has
fallen ↓↓21 Pr
,Pr
.
Real wage in terms of prices of both goods is rising.
Collectively, we conclude the wage-price analysis with the following set
of inequalities:
^^
2
^
1
^
rPPw >>> (28)
From equation (28), it can be concluded that the trade reform induced
changes in the prices of goods result in even higher changes in wages.
This is known in a more general version of S-S model as the
“magnification effect” [Jones (1965)]. The notion of a magnification effect
can have important implications for the distributional corollaries of trade
reforms. In other words, the change in commodity price of a good would
push up the price of the factor used most intensively in the production of
the commodity even more than the change in the price of commodity and
would push down the price of the other factor even lower than the change
r*
in the price of the commodity. Thus, from the above mathematical
analysis, in real terms, the labor-intensive sector gains and the capital-
intensive sector loses as a consequence of trade reforms in a developing
country. The S-S results can be noticed in Figure 6.
Fig. 6: Factor-P reflectinone good's price while holding the prices of other g
The above Figure 6 shows the impact of a
(holding prices of other goods constant)
P1=C1(w,r) and P2=C2(w,r), of two indu
intensive) intersect at point A. A price ri
due to tariff imposition—keeping P2
unchanged—leads to an outward and par
P1=C1(w,r) to P’1=C1(w,r). Thus the poin
curves of the two industries moves from A
rises from w* to w’, and return to capital (
In a more general enquiry (i.e., by relaxing
the above setting (the relationship betwee
prices) may demand addition of several
determination of factor prices. Consider
P2=C2(w,r)
r
factor oods co
chang
. Initia
stries
se of
of (c
allel sh
t of th
to B
rent) fa
the 2
n fact
more
three c
w
price (wage) due to change in
w’ g change in
w* rice Diagram
r’
P1=C1(w,r)
P’1=C1(w,r)
A
B
54
nstant. [Suranovic (2010)]
e in the price of one good
lly the unit cost curves,
(each labor and capital
(labor-intensive) good P1
apital-intensive) good 2
ift in the unit cost curve
e intersection of the two
. Consequently, the wage
lls from r* to r’.
x2x2 world assumptions),
or prices and commodity
assumptions for unique
ases where i) number of
55
goods (M) and factors (N) are equal (N=M) ii) factors are more than the
goods (N>M) or iii) factors are less than goods (N<M).
In the present study, however, the focus would be on specific H-O model
setting with 2x2x2 assumption, and more general second and third cases
are allowed for further research.
In a setting where the number of factors is equal to the number of goods,
goods and factor price relationships [equations (26-28) in this case]
become N equations with N unknowns. The question of whether factor
prices are uniquely determined by commodity prices is a matter of
whether these equations can be inverted. According to Cramer’s rule, a
linear set of equations has a unique solution if coefficient matrix A is
nonsingular, i.e., if the determinant of the matrix is nonzero.
Taking logarithm of equation (19) provides a linear set of equations that
can be expressed for N=M factors as:
∑=
=N
1jijii lnwθlnP (29)
Here Pi, i=1…..M denotes the goods prices, Wj=1…….N denotes the
factor prices (wages), and θji is the production elasticity of factor j in the
production of good i. In matrix notation the above equation can be
expressed as:
w
........θθ
......
θ..........θ
P
MNM1
1111
= (30)
Or taking the matrix equal to A:
P=Aw; N=M
The number of rows (N) is equal to the number columns (M) in the matrix.
The above square matrix would be a singular matrix if any of the two
columns or rows were proportional to each other. As it is known now that
the individual values in the above matrix are the production elasticities of
the production factors j of each good i, if any two rows or two columns are
proportional to each other, the logarithmic average cost curves of the
respective goods do not intersect each other, thus there is no unique
solution for the factor prices. This can be illustrated for the 2x2x2 case in
the factor-price diagram given below:
No
pr
ln
th
he
αβ
co
ill
ln
lnr
Fig. 7: A Linear case of 2x2x2 in Factor-Price
te that the slope of each curve is give
oduction elasticities, i.e., i
i
αβ
(solving equat
lnrαβ
αlnP
wi
i
i
i −= . The notion of proportion
at the ratios of production elasticities i
i
αβ
nce results in the same slope of
n
n
2
2
1
1
αβ
........αβ
=== . This entails that there
st curves consequently no any unique solut
ustrated in factor price diagram in Figur
P1=C1(lnw, lnr) and P2=C2(lnw, lnr) have
lnP’1=C2(lnw, lnr)
lnP2=C2(lnw, lnr)
lnP1=C1(lnw, lnr)
Diagr
n by
ion
al ro
are s
all
is no
ion f
e 7.
the
lnw
56
am
the ratio of the two
18 for one factor price:
ws or columns implies
ame for all goods, and
average cost curves.
intersection of the unit
or factor prices. This is
The unit cost curves
same slopes and run
57
parallel without intersecting. In this case the matrix P in equation (30)
cannot be inverted and cannot have a unique solution for w.
All over again, this occurs if factor intensity reversal exists. Paul A.
Samuelson (1953) shows that this would be the case if the components θji
for just two factors are proportional in two industries even if the other
factor components are not. To solve for the changes in price in an N x N
setting, we can adjust the equations (22) and (23) from a 2x2 setting as
follows:
^
NNN
^
2N2
^
1N1
^
N
^
N1N
^
212
^
111
^
1
wθ..........wθwθP
............
wθ..........wθwθP
++=
++=
Since each price change equals the weighted factor price changes
0andθθ.......θθ 1N1N1211 >+++ , an increase in one commodity price Pi,
holding the others constant must be followed by an increase in at least one
factor price wj and a decrease in another factor price r. The increase in wj
is bigger than the increase in the commodity price so that,^^
i
^
j r0Pw >>> .
Consequently, the S-S theorem (2x2x2 model) can be generalized in the
sense that a commodity price change will result in a real gain to the
abundant factor and a loss to the other factor.
Summarizing the whole discussion, it can be maintained here that factor
prices are uniquely determined for commodity prices in cases of equal
number of goods and factors or even when goods are more than the
factors, given the equation 19 is invertible (i.e., no factor reversal
intensity). This is the case when the vector of factor prices is fully
determined by the vector of commodity prices and a given commodity
price change directly infers a definite factor price change. The goods’ and
factor prices’ association is the core foundation of the empirical estimation
58
of the factor price changes due to the goods’ price change. However, S-S
Theorem does not respond to the case when factors are more than the
goods. In price elasticity terms, we can illustrate the relationship between
wages and domestic prices in following way: i
j
lnP
lnw
∂
∂
Here wj is the price of factor j and Pi is the price of good i.
4.2 Trade Liberalization, Household Demand, and Welfare Effects
This part of the Chapter will open discussion on the impact of trade
liberalization on household demand and welfare using Marshallian and
Hicksian Approaches.
4.2.1 Household Expenditure
The changes in domestic prices of traded and nontraded goods as discussed
in sections 4.1.1 and 4.1.2 owing to the changes in the trade policy affect
the households’ demands. Adjustment in households’ demands also
changes the household utility level. The aim here is to seek the minimum
level of expenditure that maintains the households’ initial utility level
when domestic prices change. To better understand the concept of
expenditure function in its entirety, consider Figure 8:
X
X
Ea
Fig
ex
iso
ex
e>
P1
res
ind
tho
Th
po
are
ea
att
low
Fig. 8: Locating the lowest level of household expenditure to attain a utility level "u"
ch of the pa
ure 8 repres
penditure to
expenditure
penditure on
0. Each expe
/P2, but dif
pectively.
ifference cu
se bundles
ough, two i
int with the
sufficient t
rlier, we are
ain a fixed u
est isoexpe
-P1/P2
rallel lines (so called isoexpe
ent all bundles of good X that
acquire given the set of
curve stands implicitly for
goods X1 and X2) for a differ
nditure function e, will there
ferent horizontal and vertica
The middle isoexpenditure
rve u(x) = u (utility level fix
of X1 and X2 where househol
soexpenditure curves (e1, e*
indifference curve u(x), e1 has
o attain the utility level u(x
interested in locating the min
tility level u(x). Clearly, that
nditure curve. And the least c
u(x)
e1/P2
e*/P2
e2/P2
2h(P,u)
Xh
nd
req
pric
1e=P
ent
fore
l i
c
ed
d yi
) h
tw
). N
imu
wi
ost
e1/P1
e*/P1 e2/P1 1h(P,u)
X2
iture curve
uire the sa
es (P1, P
1 2 2X +P X , (
level of ex
have the
ntercepts,
urve is
at u) at po
elds the sa
ave at lea
o, indicatin
evertheles
m expend
ll be at e*
bundle tha
X1
59
s) in the above
me level of total
2). Each of the
total household
penditure where
identical slope –
ei/P1 and ei/P2,
tangent to the
int Xh given all
me utility level.
st one common
g that e1 and e*
s, as mentioned
iture required to
only, i.e., at the
t achieves utility
60
u(x) at prices P1 and P2 will be the bundle h h1X =X (P,u)and h h
2X =X (P,u) . Here h
stands for “Hicksian”. If we denote the minimum expenditure necessary to
achieve required utility u(x) at prices Pi (representing set of prices P1 and
P2) by e (P, u), that level of expenditure will simply be equal to the cost of
bundle xh or h h1 1 2 2e(P,u)=P X (P,u)+P X (P,u)=e*. In more general terms,
expenditure function is expressed in following way [Mas-Colell, A. et al.
(1995)]:
n
+xÎRminp.xe(p,u)º subject to the constraint u(x) u≥ (31)
for all P values much greater than zero and all attainable utility levels u.
Note that any solution vector for this minimization problem would be
nonnegative and will depend on the parameters P and u. Also notice that if
u(x) is continuous28 and quasiconcave29, the solution will be unique, and
therefore we can denote the solution as the function hX (P ,u ) 0≥ . If this
solves the optimization problem, then the lowest expenditure necessary to
achieve utility u(x) at prices P1 and P2 will be exactly equal to the cost of
bundle h hX (P,u) or: e*P,u)=P.X (P,u) .
The above solution to the expenditure minimization problem is precisely
the consumer (household) vector of Hicksian or compensated demand
function. In fact, it is known that at a certain level of income, a change in
its given set of prices of goods will ordinarily lead to a subsequent change
in the household purchases and some corresponding change in initial level
of utility. Fixing the utility level at the initial point would help in
understanding how much an average household gains or loses from a price
change. To construct a hypothetical demand function, the whole process
must be observed by which when domestic prices fall a utility gain is
conferred on the household; it is compensated by reducing the household
28 No preference reversals 29 Balanced combination of quantities of both goods
0 2
1
income by a certain proportion (bringing it back to the initial utility level).
Similarly, whenever an increase in the price of a good is observed, causing
a utility loss, an appropriate compensation must be envisaged for this by
increasing the household income sufficiently to give a utility gain to it
equal to the loss. To get a clearer idea, we can refer to Figure 9.
The Hicksian demand curve involves constant real income and utility of
household when the domestic prices of all consumption goods change.
In the above Figur
goods X1 and X2 is
BC0 is tangent to
D
d
X10
e 9a, the
at point
the indiff
X11
C
1 X1 Fig. 9: The Hicksian and Marshallian demands for good X1 when its price P is falling
initial optim
A (X10 and X
erence curv
(b)
(a)
al household co
20) where house
e (at initial utili
P1
X01
X*
c
b
a
P11
P01
e /P
e’/P2
X20
X*2
0
X*1
X1
1
X2
X1
A
B
U0
U1
BC1
BC’1 BC
61
nsumption of two
hold budget curve
ty level U0) given
62
the prices of two goods (P1 and P2) and expenditure e0. Allowing the fall in
the price of X1 from (P10) to (P1
1), while holding the price of X2 constant,
would tilt the budget curve outward on the horizontal axis, releasing more
income for the cheaper good X1. Thus the new budget curve BC1 is now
tangent to the higher indifference curve (U1) at point B, allowing the
household to consume more of good X1. The notion of HCV reflects how
much a household is willing to pay to remain at the initial utility level
(U0). This would imply that the household should move to point C (at
initial indifference curve and not necessarily at the same consumption
bundle), which would further mean that the household is obliged to give
up the amount increased in the real expenditure in terms of good X2. This
is reflected in the downward shift of the budget curve BC1 to BC’1 and
intercepts from e*/P2 to e’/ P2 on vertical axis. Now, owing to the altered
relative prices, the households adjust their consumption to the new optimal
bundle at C (X1* and X2*).
In contrast, the idea of Equivalent Variation (EV) is how much a
household is willing to receive to reach the new indifference curve (U1)
without the change in the price of any good (compare the points B and D
in Figure 9a.) It is measured by the difference of the expenditures on the
new indifference curve (U1) in points B and D.
In Figure 9b, Marshallian (uncompensated) and Hicksian (compensated)
demand curves are driven from the indifference map in Figure 9a. Curve
(ab) is the Hicksian (compensated) demand curve and (ac) is the
Marshallian (uncompensated) demand curve in Figure 9b. The Hicksian
(compensated) demand curve shows a smaller rise in the demand for X1
because the household is obliged to pay the compensating variation to stay
at the same utility level. The demand curve depicting the Equivalent
Variation is dc in Figure 9b referring to the indifference curve (U1) in
Figure 9a. Here for a price change, the HCV is the households’
willingness to pay and Equivalent Variation is the Households’
63
willingness to receive. HCV in Figure 9b can be seen as the area under the
compensated demand curve (ab) at initial utility (U0) which is P10abP1
1.
Likewise Equivalent Variation is represented by the area under
compensated demand curve dc at new indifference curve (U1) which is
P10dcP1
1. Similarly MCS can be seen as the area under the uncompensated
demand (ac), which is P10caP1
1. Marshallian and Hicksian welfare
measures are discussed in detail in 4.2.2 and 4.2.3.
The whole construction in Figure 9b represents Hicksian and Marshallian
demand curves for good X1. Briefly, the expenditure minimization
problem is just the vector of Hicksian demands because each of the
hypothetical “budget constraints” or isoexpenditure lines the household
faces in Figure 9a involves a level of expenditure exactly equal to the
minimum level necessary at the given prices to reach the original utility
level.
Thus, mathematical expression (31) contains important information on
Hicksian demands. Repeating the definition one more time, Hicksian
demand function, also known as compensated demand, is the demand of a
consumer over a bundle of goods that minimizes their expenditure while
delivering a fixed level of utility. 30
4.2.2 Change in Marshallian Consumers’ Surplus
Furthermore, as we identified that Hicksian demand curves are not readily
observable; we would relate Hicks’ idea of Compensating Variation to the
notion of Consumer Surplus, since the later is easily measured directly
from Marshallian demand. Consumer Surplus is the amount that
consumers benefit by being able to purchase a product for a price that is 30 The discussion on the utility maximization and expenditure minimization is inspired from Mas-Colell, A. et al. (1995).
64
less than they would be willing to pay at maximum. Here it is assumed
that the Cobb-Douglas utility function describes the households’
preferences.
∫=−=0
1
P
P
00001 )dPYq(P,)Y,CS(P)Y,CS(P∆CS
(32)
or
∑∑==
−=n
1i
1i
n
1i
2i CSCS∆CS (33)
CS∆ is the change in Consumer Surplus of a household, i is the number of
goods, ∑=
n
1i
1iCS is the Consumer Surplus with actual import tariff, and
∑=
n
1i
2iCS is the Consumer Surplus with the falling general tariff in the whole
economy.
The graphical representation of MCS can be followed in the Figure 9b
presented above. The mathematical calculation of the MCS is nothing but
the measurement of the area under the uncompensated demand curve
(P01P1
1ca in Figure 9b). This can be calculated in two parts. In the first
part, the change in the estimated Marshallian demand is calculated by
multiplying the estimated Marshallian demand (Xi) with the trade reforms
induced drop in the estimated domestic prices (P01- P1
1) of selected traded
and nontraded goods. Secondly, change in the estimated Marshallian
demand (X11-X0
1) is multiplied by the trade reform induced drop in the
estimated domestic prices (P01- P1
1) of the selected goods and divided by
two, because the area under the demand curve is being measured. The
summation of the two effects i.e. 2
)X)(XP(P)P(PX
01
11
11
011
10
1i
−−+− would
lead to the measurement of the area (P01P1
1ca) or MCS with selective
protection and with liberalizing trend in general economy. This estimate is
65
then divided by the average number of households to calculate the
households’ Marshallian Consumer Surplus.
4.2.3 Change in Hicksian Compensating Variation
To measure the impact of a tariff change on household welfare, the HCV
(willingness to pay) is measured from the estimated Hicksian Demand
Equations [Varian, H. R. (2003)]. The Hicksian Demand Equations are
estimated by isolating the substitution effect from the income effect of a
price change after solving the slutsky equation in the following way:
YX
XP
XPX i
ii
Ci
i
i
∂∂
+∂∂
=∂∂
(34)
i
Ci
PX∂∂
is the substitution effect of the price change, and YX
X ii ∂∂
is the
income effect of the price change. Equation (34) can be rewritten for the
pure compensated substitution effect, i.e., the household reaction toward a
price change at the unchanged utility, in the following way:
YX
XPX
PX i
ii
i
i
Ci
∂∂
−∂∂
=∂∂
(35)
The above expression can be used to evaluate the price change in natural
log terms by first converting it into elasticity approach:
YX
YY
XXP
XP
PX
XP
PX i
ii
i
i
i
i
iCi
i
i
Ci
∂∂
−∂∂
=∂∂
or
∂∂
−∂∂
=∂∂
YX
XY
YXP
XP
PX
XP
PX i
i
ii
i
i
i
iCi
i
i
Ci (36)
66
The term on the left hand side in equation (36) is the Hicksian Price
Elasticity of Demand. The first term on the right hand side is Marshallian
Price Elasticity estimated from the Marshallian demand curves, and the
second term has two parts. First one
YXP ii is the Budget Share of each
good, and the other one
∂∂
YX
XY i
i
is the income elasticity. Thus the
equation can further be revised in following way:
−= YηXYXP
PηXPηX iii
iMii
Ci
(37)
iCi PηX is the Hicksian Price Elasticity, i
Mi PηX is the Marshallian Price
Elasticity, and YηXi is the income elasticity of demand for all goods.
Hicksian Demand Equations for selected traded and nontraded goods can
then be calculated by replacing the parameters (coefficients) in
Marshallian Demand Equations by parameters estimated here (known as
Hicksian parameters) without including the income parameter since
Hicksian demand is not a function of income but utility. Since the both
demand curves (Marshallian and Hicksian) shift in parallel by similar
amounts from their old demand curves when the prices of other goods
change, the cross price elasticities in MCS and HCV remain same since
the effect of a change in the domestic price of one good on the demand for
the other good remains same in both cases. Further, the Household
Compensating Variation can be estimated from the information on the
Hicksian household expenditures. The exercise can be performed for
actual escalating tariff and for the falling general tariff rate.
Mathematically HCV is calculated on the same lines as the MCS is
calculated in the previous section, this time Marshallian demand estimates
are replaced by the Hicksian demand estimates.
67
4.2.4 Change in the Households’ Labour Income
As per Stolper-Samuelson (S-S) theory, the free trade benefits labour-
oriented sector and harms capital oriented sectors in a developing country
setting. By the virtue of S-S theory, the workers in agriculture (labour
oriented) should gain and the workers in manufacturing (capital oriented)
should lose when the trade is liberalized. Therefore the labour income in
agriculture and in manufacturing is estimated at the domestic prices at
actual escalating tariff and at the domestic prices if the tariff had followed
the falling trend in general economy. The difference in the estimated
labour income in both sectors will help identifying the gain and the loss of
workers in both sectors if the selected commodity groups had not been
protected.
The yearly agriculture and manufacturing labour incomes31 in log linear
form are estimated using backward regression method as per following
∑=
∂∂
+=n
1ii
i
iii0i L
LQ
Pαlnlnαlnw (38)
Here wi stands for the agriculture and manufacturing labour income. The
term on the right hand side in brackets
∂∂
ii
ii L
LQ
Pln is the labour
income earned by workers in the two sectors in the production of each
good (i.e. product of the calculated domestic prices of traded goods at
actual tariff Pi, marginal productivity of each good in both sectors
i
i
LQ∂∂
and the employed labour in each sector Li). The log linear equations
for estimation of marginal productivity in the two sectors are given here;
31 The labour incomes are reported in FBS books as monthly labour income, they are converted into yearly labour incomes by multiplying with 12.
68
∑=
+=n
1iii0i lnLααlnQ (39)
and
∑=
+=n
1iji0j lnLααlnQ (40)
Here Qi and Qj are the production variables and Li and Lj are the
employment variables in the production of each selected good in
agriculture and manufacturing respectively.
Theoretically, as mentioned in earlier paragraph, the agriculture labour
income is supposed to increase and the manufacturing labour income is
supposed to decrease. The difference between the two changes (increase in
agriculture and fall in manufacturing) would be the net gain or loss to the
households in general.
Thus the total effect of the trade reforms or protection is equal to the sum
total of the Hicksian change in household welfare and the change in the
labour income (LI).
∆LICV∆Yi += (41)
Here i∆Y is the change in total household income, CV is the
Compensating Variation for the household j and ∆LI is the change in the
household’s labour income.
69
4.2.5 Change in Poorest Households’ Welfare
The above model reflects only the change in the welfare of the average
household in Pakistan if the tariff on the selected commodity groups had
followed the trend in tariff in the whole economy. It does not reveal much
explicitly about how the poorest households are affected by the
protectionist policy. Also, as mentioned earlier, trade affects poverty via
direct and indirect transmission channels. These channels, as discussed in
Chapter 2, are economic growth and changes in domestic prices and labor
income. A relative change in domestic prices of consumable goods and the
wage tends to change the household consumption possibilities.32
The most relevant question of the study, thus, is how protectionist trade
policy is affecting the poorest households’ welfare. This is answered by
computing changes in the poorest household consumption patterns when
prices of traded and nontraded goods adjust due to the change in import
tariff. Information on the calculated domestic prices of traded and
nontraded goods with actual and falling general tariff (presented in 4.1.1
and 4.1.2) is combined with statistics on the percentage of poorest
households’ monthly income spent on each selected commodity. The
detailed Table on the percentage shares of poorest monthly households’
expenditure on selected household goods is provided in Appendix A1.1.
Data on the monthly poorest household expenditure (in percent) on all
traded and nontraded goods is available in Household Income Expenditure
Surveys of various years by the Statistical Division of Pakistan. Thus the
poorest households’ demand equation is given as:
)Y,P,f(PQ PoorestNTTPoorestd = (42)
32 See Winters (2000) for a discussion on the links between international trade and poverty.
70
or
)YδPδPδδQ Poorest3NT2T10Poorestd +++= (43)
It is assumed here that the values of iδ are same for all Pakistani
households irrespective of the income level. This is justified under the
Cobb-Douglas utility setting of the Pakistani households. Here Qdpoorest
represents the quantity of a good that a poorest household demands. This
is a function of the prices of traded goods PT, the prices of nontraded
goods PNT and the monthly income of the poorest households YPoorest. 0δ is
the constant and 1δ , 2δ and 3δ are the parameters for PT, PNT and Ypoorest.
The product of the calculated domestic prices at actual and at falling
general tariff and the demand quantities of the poorest households reflect
the yearly poorest household expenditure on the selected goods in both
cases. Here it should be noted that the Cobb-Douglas utility function
depicts the poorest households’ preferences.
The yearly expenditure of the poorest household on each commodity in
money (PKR) terms is calculated by multiplying the percentage budget
share devoted for each household commodity with the poorest household
yearly income33. Further, the actual quantity of the selected goods
demanded by the poorest households is calculated from the ratio of the
yearly expenditure in PKR and the calculated domestic prices at actual and
at the general falling tariff.
The calculated poorest households’ demand equations are then used to
determine the MCS by measuring the gap between household expenditure
at actual escalating tariff and at falling tariff in general. It is done in the
similar way as the MCS for the average households estimated earlier:
33 Assumption of expenditure on any given good is a constant fraction of total household income. (Cobb-Douglas utility Function)
71
∑∑==
−≡n
1i
poorest1i
n
1i
poorest2i
poorest1 CSCS∆CS (44)
Here CSPoorest stands for the Consumer Surplus for the poorest households;
∑=
n
1i
poorest1iCS is the initial MCS for the poorest households at actual tariff and
∑=
n
1i
poorest2iCS is the MCS for the poorest households when the tariff is
following the general falling trend.
The mathematical calculation of MCS for poorest households is performed
in the similar way as for MCS for ordinary households by inserting the
calculated poorest households’ demands instead of estimated Marshallian
demands for average households.
72
5 Statistical Results and Interpretation
As discussed in Chapter three, the General Equilibrium modeling
technique does not provide empirical validation of the long time series
dataset (36 years data on domestic prices, PCI and consumption for the
demand estimation and 14 years data for measurement of households’
welfare), hence the trade-poverty link is estimated using Partial
equilibrium approach while holding the General Equilibrium framework at
the root of the analysis. The approach is adapted so since the theoretical
background of the study is based on the Stolper-Samuelson model
(extended by including the nontraded goods) which is built on the General
Equilibrium approach.
The main sources of data are FBS’ book ‘50 years of Pakistan in
Statistics’ all volumes and FBS online statistics portal34; online datasets of
Federal Board of Revenue (FBR), Pakistan, a public sector organization
formerly known as Central Board of Revenue responsible for collecting all
types of tax revenues and framing national tariff policies35; Food and
Agriculture Organization (FAO) online datasets; Household Income and
Expenditure Surveys (FBS); Labour Force Surveys (FBS) and the
International Labour Organization (ILO) online dataset. Due to the
unavailability of straight forward data on domestic and international prices
and tariff rates, the appropriate data has been calculated from available
statistics before using in the study. The detailed description on the
calculation and quality of data used in the study is provided in the
Following sections present the issues related to the quality and availability
of the data used on the domestic and international prices of all selected
goods, international trade, per capita income and the poorest household
expenditure on the selected goods. Before hand, the household food items
happen to be heterogeneous in nature therefore to avoid any ambiguity
regarding the names of the consumer goods selected for the study they are
briefly elaborated in the following Text Box. The names of the nonfood
items such as electricity, gas, firewood, cigarettes, and tea are self
explanatory, yet other food items need description with regard to their
names in the international datasets. The detailed table to identify the
selected goods at the data sources is given in Appendix A1.2.
5.1.1 Import Tariff and International and Domestic Prices
The data on the import tariff revenue from 1992 to 2005 is collected from
the Federal Board of Revenue (FBR) on various commodity groups.
Amongst, the selected commodity groups are fruits, nuts and vegetables;
Text Box: Description on domestic prices and production of selected traded and nontraded goods 1. Rice: Basmati (milled) rice 2. Milk: cow milk as cow and buffalo milk is on average more than 96% of
total milk production and total milk production. 3. Pulses: The average moong (green beans) split and washed and gram split
and the production of pulses nes (Not Elsewhere Specified). 4. Beef: Cattle and Buffalo meat 5. Mutton: Goat meat 6. Sugar: Refined Sugar 7. Vegetable Oil: Ghee, vegetable dalda tin 2.5 KGs converted into price per ton 8. Cigarettes: K2 cigarettes price per packet (converted from price per 1000
grams to price per ton and production in tons 9. Chicken meat: Chicken farm/poultry 10. Other vegetables: Vegetables fresh nes 11. Kerosene oil: Crude Oil 12. Spices: Spices nes (Not Elsewhere Specified)
74
tea, coffee and spices; milk butter and cheese; animal and vegetable oil;
edible cereals and vegetables; tobacco; fuels and oils; sugar and
confectionary; and meat, fish and other preparations. The tariff per ton on
each commodity group is calculated by dividing the total yearly tariff
revenue in PKR for 1992-2005 by the total import (C.I.F) quantity in tons
of all varieties of goods in the respective commodity group. For example,
the tariff per ton in PKR on the commodity group of meat, fish and other
preparations is calculated by dividing the total tariff revenue in PKR
collected from the commodity group by the total sum of the import
quantities in tons of all the varieties of meat (including meat, beef, chicken
meat and fish).36 As discussed in Chapter 1, the average calculated tariff
on the selected commodity groups is escalating37 in contrast to the falling
general tariff indicating a selective protectionist trade policy in the
country. The import tariff per ton on all commodities is recalculated in
line with the falling trend in the general tariff to calculate new domestic
prices to measure the loss in the households’ welfare due to selective trade
protectionist policy. The new commodity-wise tariff rates are calculated
by allowing the actual commodity-wise tariff to fall corresponding to the
general falling tariff with 1992 as the base year. The selection of the 1992
as the base year brings about zero difference between actual and the
general falling tariff and the domestic prices at both tariffs in the year
1992.
International prices are calculated by dividing the total import or export
value (in PKR) by the import or export quantities in tons for all selected
traded goods for the 1992-2005 period. There were some missing values
which are replaced by interpolation method. 36 Real world tariff rates (in percentage) are available from WTO online dataset for 1999-2002 and 2004-2005 on some goods (presented in A 4.5). These tariff rates on various goods are used as a benchmark to verify the reliability and accuracy of the average calculated tariff. The domestic prices calculated at the average calculated and real world tariff rates are nearly overlapping in some cases (pulses, milk vegetable oil, and tea) and in other cases they are better indicators of domestic prices than the estimated domestic prices at calculated tariff. See graph 10 and Appendix A1.3 37 For a discussion on import tariff on selected commodities see Chapter 1.
75
Most of the domestic prices of selected goods used in estimating the
demands are taken from FBS’ 50 years of Pakistan in Statistics (from 1970
to 1996)38 and are the averages of the prices in major cities of Pakistan.
The average variation in the domestic prices of goods across major cities
of Pakistan is trivial so these prices can best reflect the domestic prices in
Pakistan in general. [see Appendix A1.4 for the average percentage
variation in domestic prices across major cities of Pakistan]. Most of the
prices are reported in KGs and they are converted in to prices per tons.
The prices from 1997-2005 are taken from the statistical year book 2006
published by FBS. Prices per ton for Apples, Bananas, fresh vegetables
and Spices39 are the local producer price statistics taken from FAO price
datasets. The domestic prices of chicken, cigarettes, gas and electricity are
taken from the average wholesale prices40 section of FBS book. The price
of cigarettes is available in PKR per 1000 grams which is converted into
PKR per ton to synchronize with the consumption data in tons. All prices
have been taken in local currency (PKR) per ton. The calculated tariff per
ton is then added to the international prices to determine the domestic
prices at actual escalating and general falling tariff.
Further, the calculated domestic prices on average tariff rate are compared
with the estimated domestic prices at average tariff rate and the calculated
domestic prices at the real world tariff rate.41 The calculated prices at
average tariff rate are found closer to the calculated domestic prices at real
world tariff rate than the estimated domestic prices. This indicates the
methodological error in Porto (2003)42 who took the estimated domestic
prices in establishing the trade-poverty link in a developing country
setting. See Appendix A1.3 for the charts to compare the three domestic
prices of all commodities. The Figure 10 (wheat and milk estimated and 38 Volume IV. Pp. 477-503. 39 Spices are defined as the spices nes (Not Elsewhere Specified) excluding turmeric, cinnamon and black pepper 40 Volume IV. Pp. 300-437 41 Ibid 35 42 See opening paragraph of Chapter 4 for further discussion
calculated domestic prices) confirms that the average tariff rates calculated
on various commodity groups are also applicable to the individual
commodities selected in the study. Therefore the calculated domestic
prices instead of the estimated ones at calculated tariff rate are taken for
the analysis.
Fig. 10: Estimated and calculated domestic prices at averaavailable years
5.1.2 Demand estimation
The demand is estimated in linear and natural
using a long time series data on domestic
consumption43 of all selected traded and no
available for the period of 36 years from 1970
and FAO online datasets. The detailed tables
production and trade of the selected goods from
Appendices 6.1-6.5. The import and export q
production quantities of crops, primary and pro
are taken from FAO’s online production and tra
tea, sugar, fish, kerosene, gas, electricity and
43 Total consumption of selected commodities is calcsubtracting exports from the total production of the selecte
Wheat (PKR per ton)
Estimated domestic prices at calculated Tariff Calculated domestic prices at calculated Tariff Calculated domestic prices at real world tariff
Milk (PKR per ton)
76
ge and real tariff rates for the
log-linear functional form
prices, trade, PCI and
ntraded goods which are
to 2005 from FBS books
on the domestic prices,
1970 to 2005 are given in
uantities in tons and the
cessed livestock products
de statistics. Production of
firewood are reported in
ulated by adding imports and d traded goods.
77
FBS’ 50 years of Pakistan in statistics44 and statistical year book 2007 of
Pakistan. Production of cigarettes is taken from the United States
Agriculture Department in million pieces.45 The data on household total
expenditure in PKR and commodity-wise expenditure as percentage of
total expenditure and per capita income are collected from FBS.
Household surveys in Pakistan are not conducted on a yearly basis. Owing
to intermittent regime changes, the tasks of household surveys have been
suffering from big time lags. Therefore some of the segments on the
required household data, like yearly time series data on average and the
poorest households’ expenditures on selected traded and nontraded goods
do not contain complete information for 36 years. Therefore the values
for the missing years have subsequently been generated by interpolation
method.
Further, owing to the unavailability of the required data series on the
consumption of the selected traded and nontraded goods by the poorest
households in Pakistan, the poorest household demand equations are not
estimated; rather they are calculated from the available information on the
poorest households’ budget shares allocated for consumption of the
selected traded and nontraded goods, the annual household income and the
actual and general import tariff embedded domestic prices. The average
budget shares allocated to the consumption of the selected traded and
nontraded goods and the poorest households’ income and PCI are
presented in Appendix A1.1. The complete data on working hours in the
selected employment (agriculture and manufacturing) sectors for the study
period are not available. However, the available data on yearly labour
44 Sugar and tea in Volume III. p. 348 and in statistical year book 2007 p 260.; fish in volume III, p. 319 and in statistical year book 2007 p. 69; kerosene, gas and electricity in volume III, p. 328 and in statistical year book 2007 p. 155-157; firewood in volume III, p. 317. 45 Production in million pieces is converted into tons by dividing the number by 2,000,000 as the study has taken the approximate weight for one cigarette equal to 2 grams.
78
income implicitly counts for both the working hours and the
corresponding wage rate.
5.1.3 Domestic Prices and Labour Income
The detailed classification of agriculture and manufacturing workers
allocated to the production of each good separately is not available from
the labour force surveys of Pakistan. Therefore the labour employment in
the production of each good in agriculture and manufacturing is
approximated from the production shares of each good in the two sectors.
For example, the wheat production is on average only 15.49% of total
agriculture production, it is assumed that the labour employed in the
production of wheat is 15.49% of the total agriculture labour. Similarly the
production of tea is on average 0.81% of total manufacturing production,
the labour employed in the production of tea is assumed to be 0.81% of
the total manufacturing labour. By this hypothetical assumption it is
assumed that the marginal labour productivity in the selected goods is
constant which may not be true, nevertheless instead of using the total
agriculture and manufacturing labour which produces highly exaggerated
results46 on labour income generated from the production of each selected
good, this assumption can fairly correct for the exaggeration. See section
5.4.1 for a more detailed discussion on the subject in question.
Beside all the weaknesses, since all the data is collected from government
organization/ministries and international organizations, it is taken to be
unambiguously reliable for the research study.
46 The total labour income in agriculture seems to be exaggerated by the multiplication factor of 13 (number of agriculture goods) and in manufacturing by 7 (number of manufactured goods).
79
5.1.4 Exclusion of Goods from the Model
Though, intuitively, it can be argued that the liberalized trade policy
potentially can affect the whole basket of household consumption goods.
Nonetheless, due to various reasons such as data availability,
heterogeneity of household items and percentage share of household
expenditure on the items in question, some items have been excluded from
the scope of the study. Broadly, the following criteria have been adapted
to exclude the items from the analysis:
1. Item is not included because the availability of data is an issue. For
example: Rent is a nontradable good and the issue of availability of
quality data on rent remains questionable. Further, most of the
poorest in Pakistan are found staying in rural areas where the
concept of renting a house or an apartment is not as established as
it is in urban centers. It is so because almost all of the economic or
social activities such as school teaching, retail business, cropping
or other live stock activities are performed by the local residents of
the village and all villagers own their houses. Only the individuals
from government bodies such as health or education department
stay in the rural areas as outsider and mostly they are provided the
residence facilities in the same buildings where they work or the
villagers collectively arrange their stay. Secondly, Household
expenditure on rent as recorded in household surveys shows that
80% of household expenditure on rent is not explicitly spent on it
rather it is implicit. That means the amount shown under the head
of rent is not actually spent but it is the expenditure that seems to
be “saved” by poorest Households due to the “owner occupied
houses”. Other items not included here on the basis of the similar
80
presuppositions are given as under: Health, Cleaning and Laundry,
education, entertainment and recreation and Transport.
2. Item is not included because the expenditure on it as a percentage
of the total expenditure is minimal and negligible, not able to
influence the total expenditure of a poorest household
significantly. These items are: Baked and fried products, dry and
condensed milk, mustard oil, salt, eggs, coffee and other soft
drinks.
3. Item is not included because it is classified incompletely or
unclearly and is heterogeneous in nature and the collection of
individual data is too cumbersome or the good has been specified
in a very broad sense. These items are: Other cereals, other milk
products, other sugar products, ready-made food products and
other tobacco products.
On average the selection of the goods allows the present study to cover
50.77% of the total monthly household expenditure of the poorest
households (Appendix A1.1)
5.2 Regression Analysis
The statistics used in the present study originate from the national time
series data from Pakistan and no sample data is used. Therefore the
regression analysis here is performed as the descriptive statistics which
aims to summarize a data set quantitatively and portrays the relationships
among variables as they are without employing a probabilistic
formulation. In case of using sample data for regression analysis, the
researcher is confronted to inter alia two big issues which can put a
question on the reliability of the whole exercise and the results. These are:
correlation of the error terms or serial autocorrelation and different
81
variances or the heteroscedasticity. The use of heteroscedastic data does
not produce biased OLS coefficient estimates; instead it may produce
biased OLS estimates of the variance and the standard errors of the
coefficients. On the other hand, leaving unbiased OLS estimates, the serial
autocorrelation causes overestimation in the true variance of Beta,
underestimation of the estimated Beta, inflated t-stats and R2. This may
produce results looking more accurate than they actually are. Further, the
significant t-values are the indicators of the power of results estimated
from one sample to be generalized over other samples of the population.
Since, no sample data are used in the present study hence the issues arising
from the possible correlation of the error terms or different variances and
t-test significance are out of the scope. Only the large R2 values suffice the
reliability and efficiency of the regression results.47
After detailed description on the issues related to data, regression issues
and the selection of the goods in the above paragraphs, the rest of the
Chapter is divided in two parts. In the first part, the determination of the
domestic prices of traded goods at actual tariff and the falling tariff in
general is discussed. The discussion is further extended to the estimated
link of the domestic prices of traded goods with the domestic prices of
nontraded goods at actual and the falling tariff in general. Then follows the
interpretation of the impact of the trade liberalization/selective protection
on the household demand and consumption patterns for all traded and
nontraded goods via household demand equations. In the second part, the
estimation of the resulting change in the average household welfare (MCS
and HCV) is interpreted. Subsequently, the empirical results on the change
in the welfare of the poorest households in Pakistan when the tariff
follows the falling trend in the general economy are interpreted and
explicated. Finally, the estimated links between domestic prices and the
labour income in agriculture and manufacturing are interpreted. The
47 Gujarati, D. N. (2004)
82
Chapter is concluded with separate sections devoted to discussion of the
results and the limitations of the estimated interrelations.
5.3 Change in the Domestic Prices of Traded and Nontraded Goods
and Households’ Demand
The discussion on the calculated domestic prices of traded and estimated
domestic prices of nontraded goods at actual tariff and the case when the
tariff rate follows the general tariff is presented in the following section.
5.3.1 Change in the Domestic Prices of Traded Goods
The link between domestic prices and the import tariff rate given
international prices is established as per the given price-tariff
relationship P =P +ti wi i . The domestic price (Pi) of a traded good is
determined by the import tariff per ton (ti) levied on the traded good plus
the international price of the good (Pwi). Since both the determinants of the
domestic prices (import tariff originating from the trade policy of a
country and international prices determined by world demand and
supply48) are exogenous variables therefore the domestic prices of all
traded goods are calculated by inserting the values of yearly import tariff
and the international prices for the period of 1992 to 2005 in the above
price-tariff relationship. The 14-year averages of the calculated domestic
prices at actual escalating and at the falling general tariff rates given the
international prices are presented in the following Table 1. The domestic
prices calculated at actual and at falling general tariff (in PKR per ton) for
all selected traded and nontraded goods for all years are provided in
48 Pakistan being a small open economy is a price taker.
83
Appendix A3.5 and A3.6. For detailed tables on import tariff see
Appendices A4.1-4.5.
Table 1: 14-Year average domestic prices calculated at average actual and falling general import tariff rates (PKR per Ton) given average international prices and percent difference in both tariffs International
Firewood (40 KG) 20.10246 20.09384 -0.00862071 -0.043
5.3.3 Household Demand Equations and the Change in Demand for the Selected Traded and Nontraded goods
Household Income and Expenditure Survey of the statistical division of
Pakistan has included 74 household consumption goods in the household
basket. Overall 24 household goods have been selected for analysis in the
study49. The highest number from the selected goods (15) stems from the
agriculture sector. These goods are wheat, pulses, milk, butter, apples,
bananas, mutton, beef, fish, chicken, potatoes, onions, other vegetables,
49 See criteria of selection of goods in section 5.1.4
88
chilies, and other spices, followed by five agro-industrial goods, i.e.,
sugar, rice, tea, vegetable oil, and cigarettes, and four goods from the
power and energy sector, i.e., firewood, kerosene oil, gas, and electricity.
Here electricity and firewood are the two nontraded goods.
The average percentage distribution of the poorest monthly household
expenditure on selected commodities from various sectors is given in
Table 5. The detailed Table on the percentage shares of poorest monthly
household expenditure for individual commodities is presented in Appendix
A1.1.
Table 5: Percentage shares of monthly household expenditure for commodities from various sectors
Year Percentage household expenditure on various goods
Agriculture Agro-industrial Power and Energy
%
1992 32.53 11.27 5.22
1993 33.72 11.4 5.65
1994 33.72 11.4 5.65
1995 33.72 11.4 5.65
1996 30.66 12.01 4.78
1997 33.72 11.4 5.65
1998 34.85 14.85 5.28
1999 33.72 11.4 5.65
2000 33.72 11.4 5.65
2001 27.47 10.8 7.94
2002 33.72 11.4 5.65
2003 33.72 11.4 5.65
2004 38.88 13.92 6.76
2005 34.47 15.1 6.83
Average 33.72 11.40 5.65
89
It is known from Table 5 that the biggest portion of the average household
expenditure on selected goods (33.72%) is devoted to goods coming from
agriculture followed by goods from the agro-industrial sector (11.4%) and
power and energy (5.65%). Similarly the highest number of goods
consumed by the households in the country is produced in agriculture
(fifteen goods) followed by agro-industrial (five goods) and power and
energy (four goods).
The demand equations estimated for the selected goods depict the
consumption patterns vis-à-vis the domestic prices of the consumption
goods. The change in domestic prices of the goods alters the amount of the
good the households prefer to consume. With a rise in the domestic price
of a good, households adjust their demand for that good downward,
usually by substituting it with the relatively cheaper goods. Similarly,
households buy more of a good when its price in the local market falls.
Therefore the households’ welfare and the consumption possibilities the
households face change with the adjustments in the domestic prices they
pay. Hence, it is essential to investigate how the household’s consumption
arrangements of diverse goods change as the domestic prices of the goods
they consume change. This can be known from the household demand
equations for all the goods they consume. In the following section the
estimated Marshallian demand is discussed in detail.
5.3.3.1 Estimated Marshallian Demand
Marshallian demand is estimated for selected goods using OLS regression
method in linear and log linear functional forms to obtain statistically significant
estimations with a maximal R2. The yearly data on actual domestic prices of
selected goods, their substitutes’ and complementary goods’ prices including
90
applied actual tariff rates and PCI50 for 1970 to 2005 period have been included
as the independent variables. The estimated Marshallian demand equations
for the selected household goods depict a normal (negative) Price-Demand
relationship, which is in line with the conventional demand theory Q =f(P ,P ,P ,PCI)i c s for 20 out of 24 goods. (Here Q stands for quantity
demanded, Pi are the own prices of the selected traded and nontraded
goods, Pc and Ps are the prices of other (complementary and substitute)
goods. All the models maintain overall reliability of the relationships and
estimates with large values of R-square and significant F-values. In view
of the fact that the household demand equations are derived from the
national data for Pakistan and no sample data has been used, the
requirement of p-value of t-statistics below 1% in case of all models has
been compromised (see 5.2 for a brief discussion). All estimated demand
equations are provided in Appendix A2.
In the case of four goods (sugar, chilies, beef, and fish), the coefficients of
the own prices assume a positive sign. It seems most probable that the
unconventional signs of the own price coefficients in case of these goods
are the result of certain other factors causing a shift in the demand for
these goods which are beyond the scope of this study.
On the other hand, demand for all goods but pulses are positively
associated with income, which is in conformation with an average
Pakistani household consumption pattern. Pulses show a negative
association with household income; it is treated as an inferior good since
households with additional income move away from demand for pulses to
close substitutes, resulting in a fall in the demand for pulses when
50 As data on Household Income is only available for intermittent years between 1970 and 2005 the missing values in the Household Income data series are generated by interpolation method in PASW 18. However PCI is available for 36 years from 1970 to 2005. Therefore PCI has been used in estimation of Household demands to obtain statistically significant results.
91
households’ income rises. The estimated linear and natural log linear
Marshallian demand equations are given in following Table 6.
Table 6: Estimated linear and natural log linear Marshallian demand equations Demand Equations R2
Linear Demand Equations Apples Apples BananadX =26909.5-7.459P +65.625P +1.13Y 0.85
The average estimated Marshallian demand quantities for various selected
traded and nontraded goods and the resulting change is provided in the
following Table 7.
92
Table 7: Average Estimated Marshallian demand quantities in tons at actual tariff and at the falling general tariff
Selected traded and nontraded goods
Marshallian Demand quantity in tons Change in demand in tons51
Actual Tariff Case of the falling general tariff
Q1 Q2 Q2-Q1
Wheat 1,492,072.54 1,491,952.23 -120.31
Milk 961,220.05 962,021.09 801.04
Beef 147,256.45 154,126.17 6,869.72
Fish 144,148.40 147,893.23 3,744.83
Onion 47,654.11 47,904.02 249.91
Chilies 94,296.30 93,021.59 -1,274.71
Tea 2,093.45 33,28.98 1,235.54
Gas 47,329.22 47,336.89 7.66
Butter 34,481.05 34,481.05 0
Rice 4,053,707.35 4,064,581.87 10,874.52
Pulses 275,888.6 270,326.03 -5,562.57
Vegetable oil 446,232.05 452,007.50 5,775.46
Apple 265,808.52 238,749.16 -27,059.35
Banana 180,948.18 186,983.37 6,035.19
Meat 332,384.19 336,640.16 4,255.97
Chicken 151,767.52 164,502.5 12,734.98
Potato 1,693,963.26 1,691,302.75 -2,660.51
Other vegetables
1,627,168.36 1,341,447.56 -285,720.80
Other spices 27,083.09 27,254.30 171.21
Sugar 2,130,957.58 2,040,713.1 -90,244.49
Cigarettes 0 0 0
Firewood 1,157,604.31 1,155,497.36 -2,106.95
Kerosene 118,587.11 118,666.88 79.78
Electricity 67,028.64 66,694.53 -334.11
In Table 7, Q1 and Q2 are the 14-year average estimated demands for the
traded and nontraded goods at the actual and falling tariff in general.
These averages have been calculated after treating the negative demand
quantities as zero52 since the negative demand cannot be interpreted. The
change in the demanded quantity of the selected goods includes the
51 Electricity (kWh), Gas (100 cubic meter) and Kerosene (1000 ltr.) 52 Other goods having negative values for the demand quantities in some years are tea, chicken, other vegetables and Cigarettes. See Appendix A3.1 and A3.2
93
substitution and income effects of a price change. The reported rise in the
demanded quantity of the selected traded and nontraded goods is per the
expectations, however the reported average fall in the demand for wheat,
chilies, pulses, apple, potatoes, other vegetables, sugar, firewood and
electricity (shown in the light-shaded rows) need to be explained. In case
of wheat, chilies and apples their demands have reacted to the prices of
their substitutes such as rice, other spices and banana for each good
respectively. The households have partially substituted the demand for
these goods with cheaper substitutes. The demand for potatoes has
observed a fall because the households have partially substituted its
demand with onions and fall in its complementary good’s (other
vegetables) demand has caused decrease in its demand. In case of sugar, it
is expected from the positive sign of its own price coefficient that its
demand falls as its price falls. The demands for electricity has reported a
fall since its domestic price as a result of trade reform has increased and
demand for firewood has decreased since the households have substituted
it with kerosene and coal. In case of cigarettes, since its estimated demand
turns to be negative, it is treated as zero. The detailed tables on the
estimated Marshallian demand for all selected traded and nontraded goods
for all years at actual and the falling general tariff are provided in
Appendix A 3.1 and A 3.2.
5.3.3.2 Estimated Hicksian Demand
Marshallian demand functions presented in an earlier section include both
substitution and income effects of a price change on the quantity
demanded. Marshallian approach deals with cardinal utility functions
which does not distinguish between income and substitution effects of the
price change. Because the underlying utility functions of Pakistani
households are not known here therefore the Hicksian demand curves are
94
estimated. Hicksian demand function isolates the substitution effect by
postulating that the household is compensated exactly enough to buy the
combination of goods located on the same indifference curve. Therefore
Hicksian demand curves in a price-quantity space happen to be steeper
than the Marshallian demand curves because the income effect of a price
change is ignored in the former. Since the study is dealing with the gradual
year-wise changes in the domestic prices of selected goods (instead of a
big cumulative change over the course of 14 years), the deviation of
Hicksian parameters (price coefficients) from the Marshallian estimated
parameters is trivial. Further, by virtue of the fact that the Hicksian
demand of a good is determined by the domestic prices and the household
utility given the household incomes. Therefore the income parameters do
not play any role in determination of the Hicksian demand.
Therefore, the change in the Hicksian demand in the case of all traded and
nontraded goods due to trade reforms is smaller than the change in the
case of Marshallian demand. Like in the previous section on Marshallian
demands, Table 8 in this section reports the Hicksian demand at actual
tariff and the average predicted change for all traded and nontraded goods.
As expected the demand quantities under Hicksian setting are lower than
those of the Marshallian demand quantities.
Table 8: Average Estimated Hicksian demand quantities in tons at actual tariff and at the falling general tariff
Selected traded Hicksian Demand quantity in tons Change in demand in tons53
and nontraded goods Actual Tariff Case of the falling general tariff
Q1 Q2 Q2 -Q1
Wheat 1,457,047 1,456,926 -120.392
Milk 895,056.7 897,459.4 2,402.729
Beef 113,825 120,703.4 6,878.467
Fish 25,245.59 18,370.56 -6,875.02
Onion 90.96 90.00577 -0.95
53 Electricity (kWh), Gas (100 cubic meter) and Kerosene (1000 ltr.)
95
Chilies 65,954.74 64,680.01 -1,274.73
Tea 0 0 0
Gas 2,588.063 2,596.602 8.54
Butter 0 0 0
Rice 3,590,675 3,600,813 10,138.09
Pulses 479,263.4 473,702 -5,561.47
Vegetable oil 225,991.2 231,700 5,708.77
Apple 235,692.1 208,632.6 -27,059.6
Banana 92,491.86 98,525.64 6,033.776
Meat 45,291.96 48,785.15 3,493.192
Chicken 142,792.5 154,487.3 11,694.76
Potato 252,533.6 249,844.2 -2,689.35
Other vegetables 1,135,297 849,356.8 -285,940
Other spices 7,816.237 7,415.48 -400.76
Sugar 1,409,309 1,316,095 -93,214.2
Cigarettes 0 0 0
Firewood 734,909.4 732,802.4 -2,106.95
Kerosene 9,505.322 9,584.97 79.65
Electricity 32,017.03 31,682.92 -334.11
The Hicksian demand quantities are smaller than the Marshallian estimated
demand quantities. Once again, these averages of the Hicksian demand in
Table 8 are calculated after treating the negative demand quantities as zero
that may be produced by estimating the quantities on the bases of OLS-
estimates. In case of Hicksian demands, there are three goods which
reported negative estimated demand quantities during the whole period of
study (namely tea, butter and cigarettes). Further, it can be observed from
the Table 8 that now the change in the demand for two additional goods
(fish and other spices) is turning to be negative. In case of fish, the own
price elasticity is positive indicating that there can be other factors causing
a shift in its demand. As per the fish demand equation, the conventional
demand theory suggests that its demand should fall as its price falls. In the
presence of income effect in Marshallian approach the demand for fish
rose. However, under Hicksian approach, the removal of income effect
caused the decrease in the demand which is a case of fish as an absolute
96
inferior good. In case of other spices, the removal of income effect with
coefficient (0.74) almost double to its own price coefficient (-0.43) caused
a fall in its demand. See demand equations in Appendix A2. The detailed
table on estimated Hicksian demand for all selected traded and nontraded
goods for all years at actual and falling general tariff is provided in
Appendix A3.3 and A3.4.
5.4 Trade Liberalization, Household Welfare, Poorest Household Welfare and Labour incomes
This subsection is devoted to the interpretation of the empirical estimations
to investigate the effect of selective trade protectionist policy on the
household welfare. The adjustment in the welfare of the poorest households
in Pakistan owing to the trade reforms is discussed in the second part of the
section. The discussion is extended to the interpretation of the estimated
link between change in the labour incomes in agriculture and
manufacturing due to the selective protection in the last sections. Before
proceeding to the interpretations of the empirical outcomes on household
welfare, it is imperative to initiate the discussion on the issues related to the
Marshallian and Hicksian welfare measuring approaches. Following
subsection takes a detailed account on the two approaches.
5.4.1 Welfare Measuring Approaches
Welfare economics recommends consumer utility as the criterion for
measuring the consumer/household welfare. Since the utility is not readily
observable, therefore various welfare measuring criteria have always
remained controversial. Just (2004)54 has identified “not distinguishing
54 p. 98
97
between cardinality and ordinality of utility functions” as the key source
of the controversy. The cardinal system assumes that the utility can be
measured in quantitative terms i.e., in cardinal numbers such as 1, 2, 3 and
so on. Ordinal system on the other hand suggests that the utility cannot be
measured in quantitative units rather the utility drawn from different
combinations of goods can be ordered such that one is considered better,
worse or equal to the other by the consumers/households. Therefore the
utility is treated as subjective and it varies from person to person. Hence in
actual practice it cannot be measured in such quantitative or cardinal
terms. Marshallian Consumer Surplus is one welfare measure which
assumes the cardinality of the utility function. The concept of Consumer
Surplus was developed by Marshall (1930) and has formed since the basis
for most empirical welfare economic studies.55 Marshallian Consumer
Surplus has been criticized on many grounds such as its unrealistic
assumptions of independent utilities from each good,56 constant marginal
utility of money57 and interpersonal comparability58 (Mandal, 2007).59 To
cure these deficits Hicks (1939) proposes to measure people's willingness
to pay or willingness to accept to avoid a change in utility that would
follow a price change as their expression of welfare from a change of one
or of more than one price irrespective whether the underlying utility is
ordinal or cardinal. Compensating Variation is the amount of money
which, when taken away from an individual after a price change, leaves
the person just well off as before. Equivalent Variation is the amount of
money paid to an individual which, if price change does not happen,
leaves the individual just as well off as if the change had occurred. Just
55 p.6 56 This is so that Marshallian approach could not identify the case of substitute and complementary goods 57 This applies that there is no income effect of a price change and Giffen’s paradox when income effect turns negative and is stronger than the substitution effect 58 This applies that the utility of two individuals consuming the same good is comparable Just (2004) p. 4. 59 p. 14
98
(2004).60 Willig (1973, 1976) answered to the criticism on Consumer
Surplus by establishing firm theoretical relationship between two of the
Hicksian willingness to pay measures and the consumer surplus. Willig
approach suggested that Ordinary demand relationships can be used to
derive information about Hicksian compensated demands by using the
information they contain regarding income changes. Unfortunately Willig
approach is only valid for the variation of one price and for small income
changes and assumes a large error in consumer surplus if the income effect
is larger than 5% of the total price effect which is the upper bound for the
income effect of a price change. Just (2004)61. Hausman (1981) showed
how these Hicksian measures could be measured from compensated
demands using the market price and quantity data by deriving unobserved
compensated demands from observed market demands62.
Owing to the restrictive assumptions embedded in Marshallian approach
and the absence of information on the utility function of the Pakistani
households, our welfare measurement cannot only rely on the cardinal-
utility based approach. Further, since the Hicksian compensated demands
are not readily observable therefore the compensating variation is
measured from compensated demands driven from the information on
uncompensated (Marshallian) demands to measure the change in
household welfare. Slutsky equation is used to separate the income and
substitution effects from the total effect of a price change. Hausman
(1981) had the similar notion of measuring willingness to pay using
indirect utility and expenditure function.
60 p. 9 61 p.39 62 p.663
99
5.4.2 Marshallian Consumer Surplus (MCS)
Trade reforms initiate new domestic prices and consequently result in
different quantities demanded for all traded and nontraded household
consumption goods by varying proportions as elaborated in detail in the
previous sections. On average an ordinary Pakistani household loses from
the selective trade protectionist policy in the selected goods. As presented
in 4.2.4, the mathematical calculation of the MCS is the measurement of
the area under the uncompensated demand curve (P01P1
1ca in Figure 9b).
This can be calculated in two parts. In the first part, the change in the
estimated Marshallian demand is calculated by multiplying the estimated
Marshallian demand (Xi) with the trade reforms induced drop in the
estimated domestic prices (P01- P1
1) of selected traded and nontraded
goods. Secondly, change in the estimated Marshallian demand (X11-X0
1) is
multiplied by the trade reform induced drop in the estimated domestic
prices (P01- P1
1) of the selected goods and divided by two, because the area
under the demand curve is being measured. The summation of the two
effects i.e. 2
)X)(XP(P)P(PX
01
11
11
011
10
1i
−−+− would lead to the
measurement of the area (P01P1
1ca in 9b) or MCS with selective protection
and with liberalizing trend in general economy. This estimate is then
divided by the average number of households to calculate the households’
Marshallian Consumer Surplus.
The total loss under Marshallian Consumer Surplus in the all selected
goods and for all years is estimated at PKR 179.28 Billion (US
$3.82Billion)63 due to the selective protectionist policy in the selected
goods. See Table 9 for total loss over the period of 14 year in the selected
goods under Marshallian Consumer Surplus. According to the Table 9 the
selective protectionist policy has resulted in the loss in the consumers’
63 At the 14-year average exchange rate of PKR 46.885=1 US Dollar
100
welfare in all the selected goods except three goods namely wheat, other
spices and electricity. The consumers’ welfare has increased in these
goods’ consumption under Marshallian Consumer Surplus as a result of
the selective protectionist policy. This unexpected welfare gain can be
explained in the following way. In the case of wheat, since the trade
reform tended to divert the demand for wheat to its substitute (rice) it is
showing a total gain due to the selective protection. Further, as the
proposed trade reform tends to increase the domestic price of electricity so
under the protection it is showing total gain in the consumer surplus. In
case of other spices, the gain in consumer surplus has appeared from its
substitution effect on the demand for chilies. Due to proposed trade
reforms the demand for chilies has diverted to the demand for other spices.
The total loss during 1992-2005 to an average Pakistani household due to
the selective protection is equal to PKR -9179.14 or US $195.779. See
Table 9 for the summary of the results.
Table 9: Total and per household change of MCS due to the selective protectionist trade policy in the selected commodities 1992-2005 Sum of the total loss during 1992-2005
Total -179,281,617,161 -3,870,555,886 -9,179.14 -198.1765
Average -12,805,829,797.29 -276,468,277.59 -655.65 -14.16
5.4.3 Hicksian Compensating Variation (HCV)
The Hicksian approach to computing Household Compensating Variation
involves construction of the Compensated Demand Curve (also known as
Hicksian Compensated Demand Curve). The notion of HCV reveals how
the household quantity demanded varies with a change in the relative price
of a good, assuming that the household is compensated with enough
income to keep it at the initial indifference curve when the prices change.
It is nothing but the household’s willingness to pay or receive an extra
amount to offset the gain or loss of a price change. The Marshallian
approach to the measurement of household welfare, performed in the
previous section, overestimates the price effect on the household welfare,
since it includes the income effect of a price change along with the price
(substitution) effect. Thus the Hicksian approach provides an accurate
measure of household welfare, since it corrects the MCS downward by
isolating the price effect of a price change from income effect in the
Slutsky Equation. See section 5.2 for a detailed account on the comparison
of the Hicksian and Marshallian welfare measuring approaches. The
mathematical calculation of HCV has been performed in the same way as
described in 5.4.2 only after replacing the Marshallian demand estimates
with Hicksian demand estimates.
The empirical results from HCV confirm the above analysis that the
average welfare effect calculated via Marshallian approach is
overestimated by an average of approximately 9.76%, since it includes the
65 The difference between the total dollar value of losses (last row) under MCS in Tables 9 and 10 is due to the use of 14-year average exchange of US $ vis-à-vis PKR in Table 9 and actual year wise exchange rate in the Table 10. Since the interest is about measuring the household welfare loss in PKR terms, the difference in US dollar value is ignored.
103
income effect of a price change. The total national figure for the gain
under HCV is estimated at PKR 161.78 Billion (US $ 3.45 Billion) as a
result of a selective protectionist policy in the selected goods. See Table
11 for total loss over the period of 14 year in the selected goods under
Hicksian Compensating Approach. Quiet in line with the MCS, the
consumers have gained in the consumption of three goods namely wheat,
other spices and electricity. The rationale for the welfare gain in the three
goods given in 5.2.1 can be referred to here.
Table 11: Total and per household change of HCV due to the selective protectionist trade policy in the selected commodities 1992-2005
Sum of the loss during 1992-2005 Total Per Household66 Total Per household PKR US $ Wheat 8,602,432.11 0.44 183,479.41 0.01 Milk -30,290,487,679 -1,550.86 -646,059,244.5 -33.08 Beef -17,537,284,785 -897.90 -374,048,945 -19.15 Fish -14,439,906,736 -739.32 -307,985,640.1 -15.77 Onion -177,101.74 -0.01 -3,777.36 -0.0002 Chilies -440,393,619.4 -22.55 -9,393,060.03 -0.48 Tea 0 0 0 0 Gas -256,988.29 -0.013 -5,481.25 -0.0003 Butter 0 0 0 0 Rice -5,678,069,739 -290.71 -121,106,318.4 -6.20 Pulses -893,933,319.7 -45.769 -19,066,509.97 -0.98 Vegetable oil -9,949,274,701 -509.4 -212,205,923 -10.86 Apple -1,409,804,940 -72.18 -30,069,423.92 -1.54 Banana -628,574,054.8 -32.18 -13,406,719.74 -0.69 Meat -11,255,524,292 -576.28 -240,066,637.3 -12.29 Chicken -42,800,668,737 -2,191.38 -912,886,184 -46.74 Potato -297,420,699.8 -15.23 -6,343,621.62 -0.32 Other vegetables -3,029,497,127 -155.11 -64,615,487.4 -3.31 Other spices 129,233,343.8 6.62 2,756,389.97 0.14 Sugar -23,246,869,990 -1,190.23 -495,827,449.9 -25.39 Cigarettes 0 0 0 00 Firewood -97,435.5 -0.005 -2078.18 -0.0001 Kerosene -19,616,272.34 -1.0043 -418,391.22 -0.02 Electricity 15.58 0.000001 0.33 0.000000 Total -161,780,022,427.64 -8,283.07 -3,450,571,023 -176.67
The total loss in the welfare of the ordinary households during 1992-2005
is equal to PKR -8283.07 and the average yearly loss is equal to PKR -
66 As per various Household Income Expenditure Surveys the average number of households during the study period is 19531417.
104
591.65. The yearly total and average loss under HCV is presented in the
following Table 12. The detailed information on the loss in the estimated
Compensating Variation in all selected goods for all years is provided in
Appendix A5.2.
Table 12: Yearly total change in HCV and change in HCV per average Pakistani household Total Loss Loss per ordinary household
Average -11,555,715,887.57 -239,135,010.85 -591.65 -12.24
The comparative analysis of the estimations of the two household welfare
approaches suggests that both MCS and HCV approaches reflect a loss in
the household welfare due to a selective protectionist trade policy.
However, a close comparison of the two measures reveals the importance
of using both approaches instead of a single approach (MCS) in the study.
The following Table 13 presents the comparison of the estimated welfare
losses under the two approaches.
67 The difference between the total dollar value of losses (last row) under HCV in Tables 11 and 12 is due to the use of 14-year average exchange of US $ vis-à-vis PKR in Table 11 and actual year wise exchange rate in Table 12. Since the interest is about measuring the household welfare loss in PKR terms, the difference in US dollar value is ignored.
105
Table 13: Comparison of Total HCV and MCS in PKR HCV MCS % difference
Other vegetables -3,029,497,127 -4,104,470,248 35.48
Other spices 129,233,343.8 103,623,333.4 -19.82
Sugar -23,246,869,990 -34,350,221,411 47.76
Cigarettes 0 0 0
Firewood -97,435.5 -163,025.27 67.32
Kerosene -19,616,272.34 -260,983,220.4 1,230.44
Electricity 15.58 33.77 116.76
Total -161,780,022,427.64 -179,281,617,161.35
Average 9.76%
It is explained in the previous sections that the Marshallian approach
overestimates the welfare measures by the amount of the income effect in
the total price effect. In other words, if the income effect is relatively
small or negative, Hicksian measures can be larger than the Marshallian
measures. Further, the large values of the prices of the goods can also
result in the large Hicksian values. In case of pulses, the Hicksian estimate
of the welfare loss (Table 12) is larger than its Marshallian counterpart by
12.915% because the income effect of a price change on the demand for
pulses is negative (-7.63). [See Appendix A2]. It is understandable that
106
when the (negative) income effect from the total price change is excluded,
the Hicksian measure would produce a large value for the welfare loss.
Further, in case of beef, meat, chicken and other spices, the relatively large
values of their prices (per ton) cause the Hicksian measure (due to
exclusion of the income effect) to be larger than the Marshallian measures.
The domestic prices of beef at actual tariff (PKR 70,699.17 per ton), Meat
(PKR 122,005.81 per ton), Chicken (PKR 79,774.07 per ton) and other
spices (PKR 42,219.49 per ton) are relatively larger than the domestic
prices of other goods. The domestic prices of all other goods are less than
PKR 10,000 per ton except milk (PKR 24055.88 per ton), Fish (PKR
61290.72 per ton), chilies (PKR 20646.18 per ton), tea (PKR 64800.75),
Butter (PKR 50768.44 per ton), vegetable oil (PKR 40013.46 per ton) and
cigarettes (PKR 2665610.71 per ton). [See Table 1 for the domestic prices
of selected goods at actual and at the falling general tariff]. Amongst, the
HCV and MCS welfare measures cannot be compared for tea, butter and
cigarettes as the estimated demands for these goods under Hicksian
approach turn to be negative which cannot be interpreted. (See A 3.3 and
A3.4 for Hicksian demands). In addition, some goods as onion
(99337.315%), gas (2573.169%) and kerosene (1230.442%) depict large
differences between Hicksian and Marshallian welfare losses. This is
because of the exclusion of the relatively larger income effect than the
own price coefficients in the case of onion and gas [(log-linear income
coefficients in onion and gas are (0.67) and (0.516) and the price
coefficients are (-0.064) and (-0.054) respectively] and due to the large
complementary effect of the price of electricity (-5117.39) in linear
demand equation of kerosene. Another reason for the large difference
between HCV and MCS in case of onion is its negative estimated Hicksian
demand in most of the years (and treated as zero) therefore the welfare
loss under HCV is suppressed in relation to the loss under MCS.
Further, according to the above Table 13 the MCS overestimates the total
welfare loss calculated from the average values (PKR -
107
179,281,617,161.35) by 9.76% than the welfare loss under HCV
calculated from the average values (PKR -161,780,022,427.64) which
defies the upper bound of 5%, proposed in Willig (1973) to approximate
the MCS with the willingness to pay approach. Further the differences
between the Hicksian and the Marshallian losses are too large in case of
onion, gas and kerosene. [See Table 13]. Therefore the estimation of the
welfare loss under the Hicksian measure is taken to be more accurate than
the Marshallian measure and the Willig (1973) proposal of relying on
MCS as a suitable welfare measure is not accepted here.
5.4.4 Poorest Households’ Demand and Welfare (MCS)
In absence of the data on the poorest household consumption patterns,
their demand quantities have been calculated from the budget shares of the
poorest households allocated for various goods and the calculated
domestic prices (including actual and falling general tariffs separately to
obtain budget shares at both tariffs) in the following way. As presented in
4.2.5, the above setting is justified only under Cobb-Douglas utility
function which assumes that despite price changes the budget share in a
given year does not change.
YXP
S iii = (44)
Rearranging equation (44) for demand:
i
ii P
YSX = (45)
Here s is the households’ budget share (in percentage) allocated for a
good, Pi is the vector of the domestic prices (including tariff) of the goods,
108
Xi is the vector of the quantities demanded at actual and at the falling tariff
in general economy and Y is the poorest household’s Yearly income.
Quite as expected the poorest households’ demands for all selected goods
rise when the falling general tariff is applied on their domestic prices. The
calculated demands of the selected goods for the poorest households are
presented in the following Table 14.
Table 14: 14 year total sum of the poorest household demand quantities (KGs)68 of the selected goods at actual tariff and when the tariff follows the general falling tariff Q1 Q2 Q2-Q1
KG KG KG
Wheat 643.19 706.32 63.13
Milk 105.47 111.48 6.00
Beef 10.59 12.98 2.39
Fish 2.32 3.41 1.09
Onion 33.45 35.59 2.15
Chilies 11.40 11.66 0.26
Tea 7.28 8.45 1.17
Gas 0.14 0.15 0.002
Butter 2.91 2.97 0.06
Rice 69.17 75.72 6.56
Pulses 60.88 63.81 2.93
Vegetable oil 32.88 35.96 3.08
Apple 32.97 36.20 3.23
Banana 57.73 71.34 13.62
Meat 1.52 1.74 0.22
Chicken 3.86 4.69 0.83
Potato 101.41 112.97 11.56
Other vegetables 97.86 105.24 7.38
Other spices 6.35 8.47 2.12
Sugar 103.67 111.60 7.9
Cigarettes 0.46 0.56 0.1
Firewood 12,721.58 12,729.83 8.25
Kerosene 28.38 28.94 0.56
Electricity 1,072.65 1,072.58 -0.06
68 Gas (100 cubic feet), Kerosene (Liter), Electricity (kWh) and Firewood (40 KG)
109
The demand for all selected goods observes a positive change with the
tariff included domestic prices following the falling general tariff instead
of the actual tariff (see Q2-Q1 in Table 14) except electricity as its price
has increased as a result of proposed trade reform. The poorest households
demand more of all commodities (except electricity) as they perceive a
drop in the domestic prices due to tariff cuts. As in real world, the tariff
rate on the selected goods is escalating (selective protectionist policy),
therefore the predicted rise in the quantity demanded of all goods in Table
14 is indeed the loss in the poorest households’ demand due to the
protection. Following Table 15 presents the predicted loss in the single
poorest household’s welfare under MCS for all years and in all goods due
to the selective protectionist trade policy.
Table 15: Single poorest household’s change in MCS due to the selective protectionist trade policy 1992-2005 Poorest household’s loss under MCS
PKR US $69
Wheat -5,360.68 -114.34
Milk -1,973.67 -42.10
Beef -1,220.11 -26.02
Fish -222.65 -4.75
Onion -205.58 -4.38
Chilies -66.69 -1.42
Tea -815.29 -17.39
Gas -15.38 -0.33
Butter -51.70 -1.10
Rice -728.76 -15.54
Pulses -338.55 -7.22
Vegetable oil -1347.75 -28.75
Apple -207.63 -4.43
Banana -342.25 -7.30
Meat -248.59 -5.30
Chicken -630.75 -13.45
Potato -530.77 -11.32
Other vegetables -574.48 -12.25
Other spices -617.55 -13.17
Sugar -970.87 -20.71
69 Calculated at 1992-2005 average Exchange Rate
110
Cigarettes -710.34 -15.15
Firewood -8.74 -0.19
Kerosene -47.01 -1.00
Electricity 0.59 0.01
All years, all goods -17,235.2 -367.61
The total loss in the welfare of the single poorest household for the period
of 14 years (1992-2005) for all goods is equal to PKR -17,235 or US $-
367.61. The average yearly loss is equal to PKR -1,231.09. The year-wise
total loss to the poorest households in all selected goods is presented in the
following Table 16.
Table 16: Year-wise change of MCS of a single poorest household in all selected traded and nontraded goods due to protectionist trade policy
Loss in the welfare of the poorest household
Year PKR US $
1992 0.000 0.000
1993 -72.41 -2.40
1994 47.18 1.53
1995 -7.68 -0.23
1996 -136.44 -3.5
1997 -278.81 -6.5
1998 -329.21 -7.03
1999 -970.94 -18.76
2000 -1,374.56 -23.52
2001 -2,756.56 -44.87
2002 -2,817.58 -48.16
2003 -2,954.22 -51.32
2004 -3,421.42 -57.66
2005 -2,162.53 -36.15
Total -17,235.2 -298.5270
Average -1,231.09 -21.32
70 The difference between the total dollar value of losses (last row) under MCS in Tables 15 and 16 is due to the use of 14-year average exchange of US $ vis-à-vis PKR in Table 15 and actual year wise exchange rate in Table 16. Since the interest is about measuring the household welfare loss in PKR terms, the difference in US dollar value is ignored.
111
The detailed Table on the loss in the poorest households’ consumer
surplus in all traded and nontraded goods for all years due to the selective
protectionist trade policy is presented in Appendix A5.3.
5.5 Labour Incomes and the Domestic Prices of the Selected Goods
The labour income and the price relationship is estimated in log-linear for
tune to the equation 38
∂∂
+= ∑=
ii
iii
n
1i0i L
LQ
Pαlnlnαlnw in agriculture
and manufacturing. The data on the labour allocation in the production of
each good is not available in the labour force surveys. Two methods then
remain open for the estimation of the link between labour income and the
domestic prices. Either the total agriculture or manufacturing labour
should be used with the domestic prices of agriculture and manufacturing
goods in the estimation of the link. However it would produce the
exaggerated results on the labour income from the production of each
good by the number of the selected goods in each sector. Because it would
inherently be assumed that all agriculture or manufacturing labour is
allocated in the production of each single good in both sectors. Another
way is to calculate the labour allocation in the production of each good by
approximating it with the percentage share of each good in the total
agriculture or manufacturing production. If, for example, the production of
wheat is 16.54% of the total agriculture production then 16.54% of the
total labour employed in agriculture should be assumed to be allocated in
the wheat production and so on. The similar approximation may apply for
the labour employment in the production of the selected manufactured
goods. Though in this method, the underlying assumption of equal
marginal productivity of labour in the production of each good may not be
true. Nevertheless in absence of the accurate data on labour allocation in
the production of each good, this method can produce results which are
112
closer to the reality than using the total agriculture or manufacturing
employment which tend to produce exaggerated results.
Therefore the labour Li used in the above equation is the allocated labour
in the production of each individual selected good calculated by the
approximation. ilnw is the natural log of the labour income in agriculture
and manufacturing, i
i
LQ∂∂
is the marginal productivity of labour in both
sectors for agriculture and manufacturing goods, Pi the domestic prices of
the selected agriculture and manufactured goods and Li is the labour
allocated in the production of each good.
The stepwise backward regression estimates of the labour income and the
domestic prices (including actual tariff), marginal productivity and the
labour are given in the following Table 17. Since the model fit for the
agriculture labour income is perfect and R2 assumes the value of 1
therefore the influence statistics cannot be computed. The same
estimations are done for the labour income in the manufacturing sector
with domestic prices (including actual tariff) of the manufactured goods.
The empirical estimates using backward regression method for the
manufacturing labour income and the domestic prices of manufactured
goods are given in Table 18.
Table 17: Empirical link between labour incomes in agriculture and the domestic prices B (Constant) 1.346 ln_wheat_P_MP_mul_Lagri -.018 ln_milk_P_MP_mul_Lagri .005 ln_Beef_P_MP_mul_Lagri -.025 ln_fish_P_MP_mul_Lagri -.017 ln_onion_P_MP_mul_Lagri .061 ln_chilies_P_MP_mul_Lagri .088 ln_pulses_P_MP_mul_Lagri .131 ln_apple_P_MP_mul_Lagri -.004 ln_banana_P_MP_mul_Lagri .049 ln_meat_P_MP_mul_Lagri .120 ln_potato_P_MP_mul_Lagri -.016 ln_oveg_P_MP_mul_Lagri -.109 ln_other_spices_P_MP_mul_Lagri -.100
113
For the final model with dependent variable Ln_Agri_labour_income_yearly_PKR, influence statistics cannot be computed because the fit is perfect. Table 18: Empirical link between labour incomes in manufacturing and estimated domestic prices Model Unstandardized Standardized
Compare the estimated labour income in agriculture at domestic prices at
actual and the falling general tariff in the Table 19. The estimated
agriculture labour income is predicted to rise in all years except 1994 and
1998. As the year 1992 is the base year for the calculation of the tariff
rates, the difference in labour income in this year is zero (see Section
5.1.1). In 1994, the general tariff rate has increased instead of falling
which has reversed the relation between the domestic prices and wages.
The last column on the right in Table 18 reports the difference in the
estimated labour income in agriculture sector at actual and at the falling
general tariff. The negative values in this column confirm the loss in the
agriculture labour income due to the selective protection in the selected
commodities. On average the agriculture labour has suffered a yearly loss
of -2973.88 PKR71. The average labour income in agriculture is predicted
to have risen from 28643.80 PKR to 31617.68 PKR per year. In
comparison to the average yearly loss in the MCS per household (PKR
9179.14) and the loss in the HCV per household (PKR 8283), the
predicted loss in the agriculture labour income is modest. One possible
explanation for the relatively small predicted loss (or expected benefit of
the trade reforms) in agriculture labour incomes due to the proposed trade
reforms can be attributed to the highly skewed land ownership in the
country. The feudal archetype in Pakistan consists of the landlords
possessing thousands of acres of agricultural land. About 67% of
households own no land in the country. Unusually, just 0.3% households
own 55 or more acres of land across the country, suggesting a highly
skewed land ownership pattern72. The studies also argue that the skewed
land distribution results in patterns of sharecropping that exploit poor
tenants. It is a fact that most of the agriculture sector in the country is
71 The labour income in agriculture sector was predicted to increase by this amount if the selected commodities had been priced at the falling general tariff instead of the escalating actual tariff (trade reforms instead of the selective protection) 72 World Bank (2002)
115
devoid of free market principles, and the agriculture labour works under a
rigid feudal arrangement73.
On the contrary, the labour in the manufacturing sector has gained by an
average yearly amount of PKR 1390.5674 due to the selective protection in
the selected goods. Theoretically, the workers in manufacturing sector
being relatively capital oriented are supposed to lose in their labour
income when the trade reforms are implemented.
Table 20: Estimated manufacturing labour income with domestic prices at actual tariff and at domestic prices if the general falling tariff is applied
The results in the Table 20 suggest that the average labour income in
manufacturing is predicted to fall from PKR 37388.07 to PKR 35997.52
as a result of the proposed trade reforms. Thus the manufacturing labour
gained by the difference of the two labour incomes (PKR 1390.56) due to
the protection. This is in conformity with the theoretical findings of the S-
S and H-O models that the capital oriented sector loses and the labour
oriented sector gains as a result of the trade reforms in a developing
73 Malik (2005) 74 The labour income in manufacturing sector was supposed to fall by this amount if the selected commodities had been priced at the falling general tariff rate instead of the escalating actual tariff (trade reforms instead of the protection)
116
country setting and vice versa in case of protectionist policy. In the case of
Pakistan, the agriculture sector employs unskilled workers intensively
therefore gains in terms of rise in the labor incomes; the manufacturing
sector, being relatively capital intensive since it employs machinery, loses
in terms of fall in the labour income as a result of trade reforms.
Implications of the Labour income-domestic Prices Link
The major employment sectors for both categories of workers (skilled and
unskilled) reported in various Labor Force Surveys published by the
Statistics Division of Pakistan are agriculture; manufacturing;
construction; and wholesale and retail trade and hotel and restaurants.
These four sectors cover 81.13% (36 year-average) of the total employed
labor force of the country. Other sectors are mining and quarrying,
electricity, gas and water, transport, storage and communication, finance,
insurance, real estate and business services, community, social and
personal services, and activities not adequately defined, which provide a
livelihood to the remaining 19% of total employed labor force. Sector-
wise average percentage employment of the total employed labor force is
as follows: agriculture (51%), manufacturing (13%), construction (5%),
wholesale and retail trade hotel and restaurants (12%), and others (19%)
(see Figure 11). The empirical link between domestic prices and labour
incomes requires estimating (average and marginal) labor productivity in
these four sectors. However, sufficient data on household consumption
items produced in the construction and wholesale and retail trade hotel and
restaurants sectors are not readily available for the study period. And since
all of the selected household items consumed by the households originate
from the agriculture and manufacturing sectors (goods produced in agro-
industrial and energy and power sectors are assumed to belong to the
manufacturing sector), the labour income-price link has been established
between domestic prices of agriculture goods with labour incomes in
agriculture sector and domestic prices of manufacturing sector goods with
labour incomes in manufacturing sector. Since predominantly the
employed work force (64% of total employed labor) is in these two key
sectors, the labor force covering ratio is sufficiently large (64%) to
generalize, though cautiously, the results of the study on the employed
labor in other sectors.
005
Fig. 11: Average percentage sector-wise employment in Pakistan from 1970-2
117
(Various Labour Force Surveys)
Further, the agriculture sector is home to most of the unskilled workers.
Almost 39% of Pakistan’s 4.51 million unskilled workers are engaged in
productive activities in agriculture (see Figure 12). Therefore, it can be
predicted from the established estimates that a large portion of the
unskilled labor in the agriculture sector is supposed to gain from the
increase in their labour incomes, though modestly, as a result of the trade
reforms.
118
Fig. 12: Average sector-wise employment of unskilled workers in Pakistan 1970-2005 (Various Labour Force Surveys)
Further, the rise in the agriculture labour income can have a positive
spillover effect on labour incomes in construction and wholesale and retail
trade sectors because the labour income in the agriculture sector is
significantly correlated with labour incomes in construction and wholesale
and retail trade sectors. However, labour incomes in the agriculture sector
are not correlated significantly with labour incomes in the manufacturing
sector. See Table 21 for Pearson Correlation Coefficient Index for
correlation between labour incomes in the four selected sectors and
domestic prices of traded and nontraded goods.
Table 21: Inter-sector labour income correlation Agriculture Construction Wholesale
Average 1,390.56 -2,973.88 -24,790,438,019 -2,071.61
On average an ordinary household is losing net of PKR -2071.61 in its
labour income per year due to the protectionist trade policy in the selected
goods. Since the present study relies on the household’s welfare loss
measured under the Hicksian approach (from Table 11) due to the
selective trade protection (Total PKR -8283.06 for all years). Therefore,
the total effect (welfare loss) to the ordinary households due to the trade
protection would be measured by adding up the net average yearly loss in
the labour income with the average yearly loss under Hicksian approach
(PKR -591.65). Following Table 23 presents the year-wise total loss to an
75 The number of households is calculated by dividing the labour employment in the two sectors by the average size of the household (6.7) [Source: Household Income Expenditure Surveys] in Pakistan
121
ordinary household owing to the selective protectionist policy against the
general relatively liberalizing economy.
Table 23: Total yearly loss to an ordinary Pakistani household due to the selective protectionist policy
HCV loss in an ordinary Household’s welfare
Net change in an ordinary household’s labour income
Total loss to an ordinary household due to the protection
Table 22: Column: 6 Net change in LI+ HCV welfare loss
Wood, A. (1997) “Openness and Wage Inequality in Developing Countries:
The Latin American Challenge to East Asian Conventional Wisdom” The
World Bank Economic Review, 1(11).
145
Woodland, A. D. (1982) International Trade and Resource Allocation,
North-Holland, Amsterdam and New York, 1982.
World Bank (1989, 1990, 2002) Annual Reports
Yang, Y. Y. and Hwang, M. (1999) “Effects of Trade Liberalization on Domestic Prices: The Evidence from Korea 1983-1995” Department of Economics, California State University, Sacramento and Berkley (January).
146
Appendix
A1 Data and Other Descriptive Tables
A1.1 Poorest household’s budget shares on various commodities, average monthly income76 and PCI
Poorest HH Monthly Income
PCI
Wheat Milk Butter Chilies Onions Cigarettes Tea Fish Gas Beef Rice
A1.2- Identification of the selected traded and nontraded goods at various sources Goods Source Prices Production Imports Exports PKR/Ton Tons Identified as 1 Wheat Wheat flour
average quality FBS Retail Prices (1970-1996) + print out from Statistical year book 2006
Wheat (Food and Agriculture Organization Price statistics) www.fao.org/FAOSTAT
Wheat (Food and Agriculture Organization Price statistics) www.fao.org/FAOSTAT
Wheat (Food and Agriculture Organization Price statistics) www.fao.org/FAOSTAT
2 Rice Rice, Basmati Tota FBS Retail Prices (1970-1996) + print out from Statistical year book 2006
Rice, Paddy (Food and Agriculture Organization Price statistics) www.fao.org/FAOSTAT
Rice (Milled) (Food and Agriculture Organization Price statistics) www.fao.org/FAOSTAT
Rice (Milled) (Food and Agriculture Organization Price statistics) www.fao.org/FAOSTAT
packet of 20 cigarettes each packet weighing 40 grams (www.tobacco.net.au) FBS Wholesale Prices (1970-1996) + print out from Statistical year book 2006
A1.3 Estimated Domestic Prices, Calculated Domestic Prices with Average Tariff Rates and Calculated Domestic Prices with Actual Tariff Rates for Available Years
Potatoes (PKR per ton)
Banana (PKR per ton)
Apple (PKR per ton)
Vegetable Oil (PKR per ton)
Pulses (PKR per ton)
Milk (PKR per ton)
Rice (PKR per ton)
Wheat (PKR per ton)
152
stimated domestic prices at calculated Tariff alculated domestic prices at calculated Tariff alculated domestic prices at real world tariff
ECC
Fish (PKR per ton)
Chicken (PKR per ton)
Meat (PKR per ton)
Tea (PKR per ton)
Other spices (PKR per ton)
Onions (PKR per ton)
153
154
A1.4 Average percentage variation in the domestic prices of some goods in major cities of Pakistan in some years77
A 3.2 Estimated Marshallian Demands in tons for all selected traded and nontraded goods at falling general import tariff rate Wheat Milk Beef Fish Onion Chilies Tea Tons
A 3.4 Estimated Hicksian Demands in tons for all selected traded and nontraded goods at falling general import tariff rate Wheat Milk Beef Fish Onion Chilies Tea Tons
Edible prep. Cereals and vegetables tariff per 100 cubic meters81 Import (Tons) Total Revenue (PKR) tariff per ton in PKR
12.52 2,048,590.00 88,000,000.00 42.96
16.22 2,898,168.00 72,000,000.00 24.84
23.46 1,920,603.00 1,030,000,000.00 537.85
15.47 2,696,845.00 195,000,000.00 72.31
30.45 1,977,825.00 393,000,000.00 198.70
49.68 2,507,090.00 230,000,000.00 91.85
28.25 2,529,909.00 565,000,000.00 223.33
9.51 3,250,130.00 524,000,000.00 161.22
4.32 1,059,587.00 484,000,000.00 456.78
3.41 172,038.00 517,000,000.00 3,005.15
25.15 289,597.00 573,000,000.00 1,978.61
33.10 167,321.00 678,000,000.00 4,052.09
40.76 132,289.00 818,000,000.00 6,183.43
48.19 1,481,965.00 880,000,000.00 593.81 Calculation of tariff on gas (per 100 Cubic Meters) 1 million cubic feet is equal to 18.91 tons or 1 ton is equal to 52882.079 cubic feet. I cubic meter is equal to 35.314 cubic feet. Therefore 1 ton is equal to (52882.079/35.314) 1497.481 cubic meters (CUM). Example: If tariff per ton on fuels and oil in year 1992 is 178.45 PKR is same as tariff per 1497.481 CUM. Dividing the tariff per ton by 14.97481 would give us tariff per 100 cum.
81 Tariff on gas
168
A 4.3 Average tariff per ton in PKR calculated from the commodity-groups wise tariff according to the import quantity share of each individual good in the group
A 4.5 Real world tariff rate on some commodities for 1999-2002 and 2004-2005 (Average of Value Added duties in percentage)
Goat meat (Mutton)
Chicken Beef Fish Milk Potatoes Onions Vegetables Pulses Tea
1999 10 35 10 25 30 0 7.5 15 0 35
2000 10 35 10 25 30 0 7.5 15 0 35
2001 10 30 10 10 30 10 10 10 5 30
2002 10 25 10 10 25 10 10 10 5 25
2003
2004 5 25 5 10 25 10 10 10 5 20
2005 5 20 5 10 25 10 10 5 5 10
Spices Wheat Rice Vegetable oil Fruits cigarettes
20 0 15 27.91 25 35
20 0 15 27.91 25 35
20 5 10 9756.2582 30 30
20 25 10 12431.2583 25 25
82 PKR/ton 83 Ibid
172
20 25 10 12431.2584 25 25
15 10 10 12431.2585 20 25
WTO online dataset86
A5 Detailed tables on welfare loss due to selective protectionist trade policy in selected traded goods A5.1 Annual loss in MCS due to the selective protectionist policy in all selected goods against the relatively liberal trend in general in PKR Wheat Milk Beef Fish Onion Chilies Tea 1992 1,418 -532 915 772 61 214 0
A 5.2 Annual loss in HCV due to selective protectionist policy in all selected goods against the relatively liberal trend in general in PKR Wheat Milk Beef Fish Onion Chilies Tea 1992 1,403 -512 857 193 0 176 0
A5.3Annual losses to a poorest household under Marshallian Consumer Surplus approach due to the selective protectionist policy in selected traded goods against relatively liberal general trend in PKR Wheat Milk Beef Fish Onion Chilies Tea Gas
A6 Domestic Prices, Production and Trade of selected goods for 1970-2005
A6.1 Domestic prices (PKR per Ton or indicated otherwise before tariff) of traded and nontraded goods 1970-2005 Wheat Rice Pulses Milk Butter Vegetable
Cigarettes Kerosene (1000 Liters) Gas CUM87 Electricity gWh Firewood
6,000.00 68,259.89 0.00 0 0
574.00 35,214.09 0.00 0.00 0.00
1,162.00 110,117.33 0.00 0.00 0.00
678.00 232,736.26 0.01 0.00 0.00
594.00 39,820.93 0.03 0.00 0.00
3,754.00 29,054.84 0.02 0.00 0.00
5,709.00 71,043.92 0.03 0.00 0.00
4,390.00 71,132.48 0.05 0.00 0.00
3,607.00 73,581.67 0.05 0.00 0.00
1,650.00 176,648.86 0.07 0.00 0.00
2,562.00 110,689.39 0.10 0.00 0.00
991.00 0.00 0.06 0.00 0.00
909.00 16.47 0.00 0.00 0.00
908.00 66.45 0.01 0.00 0.00
690.00 0.00 0.00 0.00 0.00
584.00 0.00 0.00 0.00 0.00
582.00 0.00 0.00 0.00 0.00
518.00 0.00 0.00 0.00 0.00
411.00 123,020.28 0.00 0.00 0.00
286.00 221,364.83 0.00 0.00 0.00
1,464.00 336,287.52 0.00 0.00 0.00
1,998.00 352,357.36 0.00 0.00 0.00
2,469.00 444,478.96 0.00 0.00 0.00
1,171.00 316,748.88 0.00 0.00 0.00
981.00 377,164.34 0.00 0.00 0.00
1,066.00 382,854.12 0.00 0.00 0.00
1,329.00 1,056,145.62 0.00 0.00 0.00
1,226.00 163,013.64 0.00 0.00 0.00
1,642.00 445,124.97 0.00 0.00 0.00
978.00 460,856.45 0.00 0.00 0.00
1,109.00 473,518.96 0.00 0.00 0.00
1,202.00 446,276.05 13.59 0.00 0.00
1,408.00 254,042.62 1.52 0.00 0.00
1,384.00 147,453.19 2.08 0.00 0.00
1,019.00 58,150.75 2.28 0.00 0.00
87 Trade statistics of Kerosene and Gas in (1000 liters and CUM) have been calculated from International prices of crude oil per liter and Average international (1985-2009) price of natural gas per cubic meter and the value of imports and exports in US $.
188
623.00 0.00 3.38 0.00 0.00
A 6.5 International Prices of crude oil (US $ per Barrel) and natural gas (PKR per CUM)
International Crude Oil
Price US$/Barrel88
1970 4.1
1971 4.2
1972 4
1973 4.9
1974 36.2
1975 47.5
1976 38.9
1977 37.1
1978 35.1
1979 47.10
1980 62.5
1981 71.8
1982 61.9
1983 49.3
1984 47.60
1985 43.20 182.96
1986 22.80 147.43
1987 26.00 121.20
1988 20.10 124.89
1989 24.10 137.86
1990 28.40 153.60
1991 25.00 160.53
1992 22.50 169.60
1993 20.80 216.21
1994 18.00 203.72
1995 19.00 176.64
1996 23.30 271.25
1997 22.90 330.97
1998 13.80 325.51
1999 17.90 409.52
2000 30.00 815.18
2001 27.00 961.02 2002 27.00 650.75
88 1 US Barrel of oil is equal to 160 Liters.
189
2003 30.00 1051.07
2004 38.30 1217.50
2005 50.10 1715.62
Average 632.62
CV
1. General
Name: Shaikh, Naveed Ahmed, PhD Date of birth: 02.01.1976 Nationality: Pakistan Civil Status: Married Contact: Office: Department of Economics, Shah Abdul Latif University, Khairpur, Pakistan Ph. 0092-243-9280280 Home: 37/192 Saddar Mohalla Shikarpur Sindh Cell: 0092-336-2401093 Email: [email protected]
2. Professional Experience (11 years)
September 2000 to December 2005 Lecturer (Department of EcoShah Abdul Latif University, Khairpur, Pakistan)
December 2005 to date... Assistant Professor (Departm
Economics, Shah Abdul LatifUniversity, Khairpur, Pakistan
3. Academic Record
Degree/Certificate Year University
• PhD 2011 Institute of dev(International Development Studies) Research and
Development PRuhr UniversityBochum, Germ
Thesis Topic: Trade Liberalization, Poverty and Welfare in Pakistan
Supervisor: Prof. Dr. Wilhelm Löwenstein
• M.Phil (Economics) 2003 Department of EcShah Abdul Latif University, KhairPakistan
Thesis Topic: Information Technology as a catalyst agent for the process of Globalization (A case study of Pakistan)
Supervisor: Prof. Dr. Iqbal Ahmed Panhwar
• MSc Economics 1998-1999 School of Economics, International Islamic University, Islamabad, Pakistan
Cumulative Grade Percentage Average (CGPA): 3.4/4.00 equal to 75.9%
• BSc (Hons) Economics 1994-1998 School of Economics, International Islamic University, Islamabad, Pakistan
Cumulative Grade Percentage Average (CGPA): 2.89/4.00 equal to 70.55%.
• Intermediate (Pre-Eng.) (69%) 1993 C & S Government Degree College Shikarpur
(Board of Intermediate and Secondary Education Larkana)
• Matriculation (84%) 1991 Government High School No.1 Shikarpur
(Board of Intermediate and Secondary Education Larkana)
Remark: English was the major language of instruction throughout the academic career.
4. Conferences, Internships, Trainings, Workshops and Projects
Conferences
• Abstract accepted titled “Selective Trade Protectionism and Trade Liberalization: Impact on Household Welfare in Pakistan (A Marshallian Approach)” for 27th PSDE-PIDE annual conference at Pakistan Society of Development Economists, arranged by Pakistan Institute of Development Economics, Islamabad 16-17 December 2011.
192
• Abstract accepted titled “Trade Liberalization and Wage Inequality: Conflict of Statistical Evidence in the Reviewed Literature” for SZABIST’s International Conference on Management, Social Sciences, Economics and Computing in collaboration with the Faculty of Administrative and Management Sciences, University of Karachi.14-15 December 2011.
Internships
• Three month internship at State Bank of Pakistan (Central Bank) to get the know how about the working of its various sections (2 July 1998-26 August 1998)
Trainings
• 3-month short Course on International Trade, WTO and related Issues at Pakistan Institute of Development Economics, Islamabad, Pakistan ( June - August 2003)
• One year Diploma in Information Technology at Petroman Training Institute Sukkur, Pakistan ( July 2000-June 2001)
Workshops
• Three day Workshop on “Intercultural Communication and Team Building” organized by Institute of Development Research and Development Policy, RUB (15-17 September 2006 at Olpe, Germany )
• “Narration/Mediation: The Constitution of the ‘Self’ in Interdisciplinary Perspective” Section Day Humanities and Social Sciences (November 7, 2008) arranged by Ruhr Research School, Ruhr University Bochum, Germany
• Three day writing Workshop on “Becoming a Better Academic Writer” (14-16 May 2009) at Ruhr Research School, Ruhr University Bochum, Germany.
• Three day Workshop on “Introduction into Structural Equation Modeling using Mplus at Institute of Development Research and Development Policy, Ruhr University Bochum, Germany (17-19 February 2010)
• 3-day Reintegration Seminar arranged by ARBEITSKREIS AFRIKANISCH-ASIATISCHER AKADEMIKERINNEN UND AKADEMIKER Göttingen, Germany (23-25 October 2009)
• 3-Day workshop on Project Proposal writing at Institute for Development Research and Development Policy, Ruhr University
193
Bochum in collaboration with GOPA Consultants. (10-12 March 2010) (Certificate awaited…)
• 2 day workshop on “Social Audit of Governance and Public Service Delivery 2011-2012” by UNDP at Holiday (Islamabad) Hotel, 10-11 January, 2012
Projects
• A Household Survey completed successfully as Research and Logistic Coordinator under the project of Social Audit and Government Services Delivery 2011-2012 under Partnership between UNDP and Shah Abdul Latif University Khairpur funded by UNDP (Amount PKR796500) in January-February 2012. Conducted two-day class and field training of 9 team members and 8 days field survey in four different districts of Khairpur, Sukkur, Shikarpur and Jacobabad. (Amount Rs. 0.8 Million)
• Proposal accepted for Gender Equity Project funded by USAID in partnership with Asia Foundation (Amount Rs. 3 Million)
5. Weakly Business and Finance Review articles: 1. US dollar in the line of fire
Link to access: http://jang.com.pk/thenews/jan2010-weekly/busrev-18-01-2010/p6.htm
2. INTERNATIONAL TRADE : Lessons from experience Link to access: http://jang.com.pk/thenews/feb2010-weekly/busrev-01-02-2010/p5.htm
3. Challenges the new economic advisor is going to face Link to access: http://jang.com.pk/thenews/mar2010-weekly/busrev-29-03-2010/p6.htm
4. Economic Slowdown and rising poverty Link to access: http://jang.com.pk/thenews/may2010-weekly/busrev-10-05-2010/p13.htm
5. Global Outlook: Greek Financial Crisis: Another disaster unleashed Link to access: http://jang.com.pk/thenews/may2010-weekly/busrev-17-05-2010/p4.htm
6. Shift in Global Economic Power Link to access: http://jang.com.pk/thenews/may2010-weekly/busrev-31-05-2010/p4.htm
7. An assessment of Revenues and Expenditures Link to access: http://jang.com.pk/thenews/jun2010-weekly/busrev-14-06-2010/p9.htm
8. Pak-Iran Gas Pipe Line Link to access: http://jang.com.pk/thenews/jul2010-weekly/busrev-12-07-2010/p4.htm
Link to access: http://jang.com.pk/thenews/jul2010-weekly/busrev-26-07-2010/index.html#1
6. Research Publications
1. Shaikh, N. A. & Memon, A.L, (2000); “Credit Creation in general & Banking Practice in Pakistan” The Commerce & Economic Review, Vol. XI 2000-01, Shah Abdul Latif University Khairpur, Pakistan
2. Shaikh, N. A. & Memon, A.L, (2002); “Impact of war against terrorism on Pakistan Economy” The Commerce & Economic Review, Vol. XII 2002-03, Shah Abdul Latif University Khairpur, Pakistan
3. Shaikh, N. A. & Sahito, I. H. Dr, (2002) “Role of Microcredit in Economic Revival and Poverty Alleviation” The Commerce & Economic Review, Vol. XII 2002-03, Shah Abdul Latif University Khairpur, Pakistan
4. Jamali, M. B. Dr.; H. Jawad, Shaikh, N. A. Dr; Shaikh, F. M, Afridi, T. (2011) “Internationalization of SMES and Organizational factors in developing countries: A case study of ice industry in Pakistan” Australian Journal of Business and Management Research Vol. 1, No. 7 [129-138] | October
5. Shaikh, N. A. Dr.; Mangi, R. A.; Soomro, H. J.; (2011) “Trade Liberalization and Wage Inequality: Conflict of Statistical Evidence in the Reviewed Literature: Experience of Latin American and East Asian Countries corresponding to the theoretical findings of Heckscher-Ohlin Trade Theorem” Interdisciplinary Journal of Contemporary Research in Business, Vol. 3, No.8, Dec. [965-971]
6. Bhatti, N., Maitlo, G. M., Shaikh, N. A. Dr., Hashmi, M. A., Shaikh, F. M., (2012) “The Impact of Autocratic and Democratic Leadership Style on Job Satisfaction” International Business Review, Vol. 5, No. 2; Feb.[192-201] www.ccsenet.org/ibr .
7. Bhatti, N., Phulpoto, L.A, Shaikh, N.A. Dr, Afridi, T., Shaikh, F.M (2012) “Economic and Social Factors of Poverty: A case study of Sindh” Journal of Management and Sustainability, Vol. 2 No.1, March.[227-234]
8. Shaikh, S.; Shaikh, N.A. Dr. (2012) “Impact of FDI, Capital Formation and International Trade on Economic Growth of Pakistan: An Empirical Analysis” Interdisciplinary Journal of Contemporary Research in Business, Vol. 3, No. 11, March.