economic determinants of international remittances...
TRANSCRIPT
ECONOMIC DETERMINANTS OF INTERNATIONAL REMITTANCES IN PAKISTAN
By
Nisar Ahmad
A Thesis Submitted to the Department of Economics
In Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy in Economics
DEPARTMENT OF ECONOMICS UNIVERSITY OF SARGODHA
SARGODHA 2008
i
CERTIFICATE
This thesis “Economic Determinants of international Remittances in Pakistan” by
Nisar Ahmad is accepted in its present form by the Department of Economics as
satisfying the thesis requirement for the degree of Doctor of Philosophy in
Economics.
Supervisor -----------------------------
Chairman ----------------------------
External Examiner ----------------------------
ii
CERTIFICATE
It is certified that Mr. Nisar Ahmad, Ph.D Scholar (Reg. UOS/Ph-D/Eco/05/05)
has taken his M.Phil degree from BZU, Multan, which is recognized as
equivalent to the degree of M. Phil of the University of Sargodha.
Moreover, it is certified that for not less than three years after passing his
examination for the master’s Degree, he has pursued an Advance course of
Study and Research in Economics for Ph.D Degree.
Chairman, Department of Economics University of Sargodha Sargodha
iii
DECLARATION
I, Nisar Ahmad, Ph.D Scholar, Registration No UOS/Ph-D/Eco/05/05, University
of Sargodha, hereby declare that the research work (Ph.D Dissertation) which I
have submitted has not already been submitted and shall not in future be
submitted for obtaining similar degree from any other University.
Nisar Ahmad PhD Scholar, University of Sargodha, Sargodha.
iv
DEDICATED TO MY LOVING SONS
ABDULLAH NISAR
MUHAMMAD SHOAIB NISAR
MUHAMMAD USMAN NISAR
MUHAMMAD JUNAID NISAR
v
ACKNOWLEDGEMENT
All thanks and praises to ALMIGHTY ALLAH (SWTH), the Merciful, the
Compassionate, who provided me the opportunity and gave me the strength to
complete this research work within stipulated time and all praises for HIS HOLLY
PROPHET HAZARAT MUHAMMAD (P.B.U.H), the city of knowledge and torch
of guidance for humanity as a whole.
I would like to express my gratitude to my supervisor and the worthy chairman;
department of economics, Dr. Zakir Hussain. I am extremely indebted to him for
his kindness. It is all due to his full co-operation from beginning to end that I
have completed this research study successfully. I learnt the art of hard work
from the personality of my supervisor. The foreign faculty was a great blessing of
ALLAH (SWTH) for the Department of Economics, University of Sargodha,
Sargodha, Therefore, I also like to acknowledge the services and guidance
provided by the foreign faculty professors Dr. Maqbool H. sial, Dr. S.J. Malik, and
Dr. Khalid Raiz. I thank them for their co-operation and proper guidance during
the research work.
I would like to thank the various researchers and economists for their contribution
toward writing valuable research material about the different aspects of
remittances and migration. I recognize the contribution of Federal Bureau of
Statistics, Islamabad and State Bank of Pakistan, Karachi for publishing data. I
vi
also recognize the support of Higher Education Commission for providing Ph.D
indigenous scholarship to accomplish this task.
I am thankful to my father, brothers, relatives, friends, and colleagues who
encouraged me for completing this research work and especially to Mr. Nadeem
Sohail, Mr. Haroon Sarwar, Mr. Naveed Iqbal and Mr. Adeel Saleem, Ph.D
students. I am greatly indebted to Dr. Masood Sarwar and Dr. Abdul Saboor for
explaining econometric techniques for the estimation. I owe a word of thanks to
all faculty members of Department of Economics, University of Sargodha for their
moral supports. I have special word of thanks and appreciation for Mr. Khurram
and Mr. Imran who assisted me in composing the research study.
Last but not the least I am grateful to my wife Rukhsana Kusar for her assistance
and encouragement in difficult times. She accepted all the family responsibilities
and provided me every support to complete the research work.
NISAR AHMAD
vii
CONTENTS
CH # NAME OF CHAPTER PAGE No
ABSTRACT 1
1. INTRODUCTION 2
1.1 Introduction 2
1.2 Remittances in Pakistan 7
1.3 Objectives of the Study 9
1.4 Plan of Study 10
2. REVIEW OF LITERATURE 11
2.1 Introduction 11
2.2 Literature review at National Level 12
2.3 Literature review at International Level 19
2.4 Literature review about Remittances at Micro Level 19
2.5 Literature review about Remittances at Macro Level 27
2.6 Literature review about Migration 44
2.7 Summary 54
viii
3. THEORTICAL FRAMEWORK 57
3.1 Introduction 57
3.2 Theories of Migration 58
3.2.1 Neo Classical Theory of Migration 59
3.2.2 Push- Pull Theory of Migration 61
3.2.3 Dual Labour Market Theory of Migration 61
3.2.4 Migration Systems Theory 63
3.2.5 The New Economics of Labour Migration 64
3.2.6 Trans-nationalism Theory of Migration 64
3.2.7 Roy’s Theory of Migration 65
3.3 Theory of Remittances 69
3.4 Motives of Remittances 71
3.5 Hypotheses Formulation 76
4. DATA AND METHODOLOGY 78
4.1 Introduction 78
4.2 Source of Data 79
4.3 Methods and Procedures 80
4.4 Stationary, non-stationary and Order of Integration 82
4.5 The concept of Co-integration 83
4.6 Order of Integration 84
ix
4.7 Test of Co-integration 86
4.8 Co-integrating Vectors 87
4.9 Error Correction Mechanism 88
4.10 Modeling Strategy 91
5. UNIT ROOT TEST 94
5.1 Introduction 94
5.2 Unit Root Test 95
5.3 Graphical Representation of the Series 101
6. COINTEGRATION ANALYSIS 110
6.1 Introduction 110
6.2 Co-integration 110
6.3 Co-integration Analysis of Remittances Model 111
6.4 Co-integration Analysis of Migration Model 116
7. Economic Determinants of International Remittances 121
7.1 Introduction 121
7.2 Long Run Relationship 121
7.3 Error Correction Mechanism 126
7.4 Long Run and Short Run Elasticities 128
x
7.5 Elementary Analysis of Workers’ Remittances in Pakistan 130
7.6 Impulse Response Function of Remittances Model 133
8. Economic Determinants of International Migration 136
8.1 Introduction 136
8.2 Long Run Relationship 137
8.3 Error Correction Mechanism 139
8.4 Long Run and Short Run Elasticities 141
8.5 Impulse Response Function of Migration Model 142
9. CONCLUSION AND POLICY RECOMMENDATIONS 145
REFERENCES 153
APPENDIX 162
LIST OF ABBREVIATIONS 168
xi
LIST OF TABLES
T. NO. NAME OF TABLE PAGE NO
5.1 Unit Root tests using Augmented Dickey-Fuller Method 97
5.2 Unit Root tests using Augmented Dickey-Fuller Method 98
5.3 Unit Root tests using Augmented Dickey-Fuller Method 99
5.4 Unit Root tests using Augmented Dickey-Fuller Method 100
6.1 Selecting the Order of the VAR for Real Remittances Model 113
6.2 Johansen Co-integration Results for Real Remittances Model 115
6.3 Johansen Co-integration Results for Real Remittances Model 116 6.4 Selecting the Order of the VAR for Migration Model 118
6.5 Johansen Co-integration Results for Migration Model 119
6.6 Johansen Co-integration Results for Migration Model 120
7.1 Regression results relating Real Remittances with Independent
Variable in Pakistan 123
7.2 Regression results relating First Difference of Real Remittances With First Difference of Independent Variables in Pakistan 127
7.3 Long Run and Short Run Elasticities of Remittances in Pakistan 129 7.4 Generalized Impulse Response(s) to one S.E. shock in the
Equation for LRREM 134
8.1 Regression results relating Migration with Independent Variables in Pakistan 138
8.2 Regression results relating First Difference of Migration With
First Difference of Independent Variables in Pakistan 140
8.3 Long Run and Short Run Elasticities of Migration in Pakistan 142
xii
8.4 Generalized Impulse Response(s) to one S.E. shock in the Equation for LNW 143
A1 Country Wise Remittances ($ Million) in Pakistan 162 A2 Country Wise Percentage Share of Remittances in Pakistan 164 A3 Remittances as Percentage of GDP, Exports, Imports and Savings 166
xiii
LIST OF FIGURES
F. NO. NAME OF FIGURE PAGE NO
5.1 Real Remittances in Pakistan 101 5.2 Real GDP in Pakistan 102 5.3 Real Growth Rate in Pakistan 103 5.4 Real Wage Rate in Pakistan 104 5.5 Unemployment Rate in Pakistan 105 5.6 Literacy Rate in Pakistan 106 5.7 Bank Spread Rate in Pakistan 107 5.8 Out Flow of Migrant Workers per year from Pakistan 108 5.9 Inflation Rate in Pakistan 109 7.1 Generalized Impulse Response (s) for Remittances Model 135 8.1 Generalized Impulse Response (s) for Migration Model 144
1
Abstract
The international remittance is one of the important sources of foreign exchange
earnings in the country. Recently, Pakistan received over $4 billion as foreign
remittances in 2006. The flows of funds through foreign remittances served as a
cushion for balance of payment and foreign reserves. The economic
determinants of international workers’ remittances and migration in Pakistan
were investigated in this research study. The time series data were used to
identify the economic determinants of the international remittances and migration
in Pakistan for the year of 1973 through 2005. The stationary properties of each
time series were investigated and found integrated of order one. The variables of
remittances and migration models were established as co-integrated.
The real remittances in Pakistan were found positively related with real GDP, real
Growth rate and unemployment rate and were negatively related with real wage
rate, literacy rate and spread rate of banks in Pakistan. The migration from
Pakistan was found positively related with real remittances, inflation and
unemployment rate and was negatively related with real wage rate in the country.
Based on research results, it seems imperative to introduce financial innovations
and create friendly environment for migrant workers to invest in the country. The
widening gap of saving-investment can also be bridged through providing
incentives on the bank rate and saving schemes.
2
Chapter 1
Introduction
1.1 Introduction
The remittances were generally defined as a part of the earnings of the migrants,
who were working in other parts of country or foreign countries. The worker may
move from one place to other places for their livelihood. The remittances were
termed as internal remittances when sent from one region to other within the
country and were known as external or international remittances when sent from
one country to other country. The workers may have migrated to different places
within their own countries, or left their countries of origin completely toward
foreign countries for the sake of more earnings. Precisely, international
remittances were the working migrant’s earnings sent back from the foreign
country to their home country in which their families were residing. Hence
international remittances were the proportion of the earned income of the migrant
workers sent to their families at home country.
The remittances may be classified into different types and groups as for example
internal remittances and external (international) remittances. Here, in this study,
the concern was about the international remittances those were further divided
into four groups (Wahba, 1991). These four types of international remittances
were: (i) Potential remittances were the amount of the savings from the total
3
earning of the worker living in the foreign country. Therefore, such type of
remittances might be the maximum amount of remittances that migrant was able
to send to their families at home. (ii) Fixed remittances were the remittances sent
to the family of migrant worker for the fulfillment of their needs including food,
clothing, shelter and education etc. Therefore, such type of remittances showed
the minimum amount of remittances transferred to the home country. (iii)
Discretionary remittances were amount of remittances transferred in excess of
fixed remittances. Therefore, these amounts of remittances were the actual level
of remittances coupled with fixed remittances. (iv) Saved remittances showed the
difference between potential remittances and the amount of remittances
transferred to home country during the prescribed period. The flows of
remittances were accumulated into a stock of resources of the home country,
which were used to supplement actual remittances at a later date. The stock of
wealth was the result of a portfolio decision by the migrant workers and these
workers were encouraged to contribute their resources for the development of
their home country. The types and groups of remittances might be highly
important due to their different components were driven by varying motives
(Wahba, 1991).
The transfer of international remittances was linked with the pattern of
international migration and mobility of the labor. Therefore remittances and
migration were a key source of development and income redistribution at national
as well as at international level. International migration and remittances have a
4
pivot role in generating and redistributing income among different countries
during the nineteenth century (Taylor, 1994). Given the sizeable productivity
differences among countries, the size of the welfare achieved from international
migration might be very large according to productivity sizes of the different
countries. It was estimated that the gains from international migration in world
output was ranged between from 20 per cent to 40 per cent of world output
(Hamilton and Whalley, 1984).
Over the time, the remittances were increasing rapidly and played important a
role in the development of different countries all over the world. The volume of
remittances has jumped from $58 billion to $160 billion during the time period of
1995 to 2004. Perhaps, this volume has crossed private debt, equity flows and
official assistance for the development purposes. In this way, remittances might
be ranked next to foreign direct investment in the developing countries. The
number of factors was responsible for the increase in the volume and growth of
remittances including the rise in the supply of migrants, the fall in transaction
costs of migration etc. The concerted efforts of receiving remittances
governments were about strengthening, monitoring and data gathering
capabilities to gauge the remittances in their countries. However, fall in
remittances was associated with the increasing preference of immigration
policies for the admission of high skilled workers (World Bank, 2006).
5
The volume of remittances was reduced in Africa and countries on the southern
shore of the Mediterranean due to declined migration rate witnessed in these
regions during the period from 1990 to 2000. However, the share of remittances
in GDP was increased for Latin America due to its growing integration into the
world economy (Docquier and Marfouk, 2004).
It was observed that the remittances bring benefits as well as socio-economic
cost to the society. The people spent their remittances on conspicuous
consumption; this act of unnecessary consumption caused inflation in the
remittances receiving countries and pushed wages upward. The consumer
demanded imported items; ultimately, it increases dependency upon imports.
Therefore, the problems of balance of payment (BOP) started to be generated in
the country. The investment in the productive projects in the country started to be
slow downed. The problems of brain drains and moral hazards in the developing
countries were also seen due to remittances (Buch and Kuckulenz, 2002)
It was seen that the economic workers in the foreign countries spend their
earned income on unnecessary consumption. Moreover, the migrant workers
were not provided enabling environment for productive investment. The analysis
of remittances at micro and macro level seems imperative for planner and policy
makers. As a result, workers’ remittances have gained increasing interest from
both researchers and policy makers over the last years. The micro and macro
factors entailed important bearing on the flow of remittances. The important
6
microeconomic determinants of international remittances in literature were
education and income level of the migrant, the length of the stay of the migrant
worker in the host country, the number of the children at home, and the marital
status of the migrant (Swamy, 1981, Lucas and Stark, 1985). The
macroeconomic determinants of remittances included a number of factors such
as: the number of workers, wage rates, economic opportunities available in host
as well as in the home countries, situation of inflation in both countries, spread
interest rate differentials in these countries, currency exchange rates between
home and host countries, the level of political stability in home as well as in the
host country, and facility of transferring funds from host country to home country.
The government immigration policies were also considered very important factor
of the volume of the international remittances in any country (Russell, 1986,
Sakka and Mcnabb, 1999).
The volume of remittances received in the developing countries might improve
the growth and development pattern of these countries and reduce poverty level
in these countries. Remittances could maintain macroeconomic stability in these
countries and reduce the impact of adverse shocks in these countries. The poor
people in these countries mainly depended upon the amount of remittances for
their basic needs including food, shelter and health; therefore, the increased
remittances could reduce poverty and inequality in the country. The construction
activity was found highly related with remittances inflow empirically (IMF, 2005).
7
1.2 Remittances in Pakistan
The workers remittances in Pakistan continued to be a significant component of
balance of payments and have made tremendous contribution towards overall
foreign exchange earnings of the country. The inflow of workers remittances in
Pakistan was widely fluctuating over time. The amount of remittances in Pakistan
reached to $2.88 billion in 1982-83 with continuous increasing trend. The pattern
of remittances in Pakistan started fluctuating after the period of 1982-83. The
volume of remittances has declined from $1.84 billion in 1990-91 to $1.40 billion
in 1996-97. However, the remittances again slightly rose to $1.48 billion during
1997-98 and fell to $0.98 billion in 1999-2000. In the period of 2002-03, the
volume of remittances touched to a maximum of $4.23 billion and again came
down to $3.87 billion in 2003-04 (SBP, 2004).
The number worker have been migrated to the Middle East during the period
from 1970 to 1985.This pattern of migration to the Middle East has unique
characteristics in number of ways. The young males were the principal migrants
to Middle East without the migration of their whole families, their families were
stayed in Pakistan and migrant workers sent a bulk of remittances to their
families in Pakistan. Secondly, the majorities of migrant workers were unskilled
and belong to low-income households in Pakistan. Therefore, the remittances
were a windfall for these low-income households. These families were now able
to set up small businesses, acquire real estate and make substantial
improvements in their living standard. The construction activities were gone
8
slowly down in the Middle East in the beginning of 1990s. After 1990s, there
were less earning opportunities in the Middle East for unskilled Pakistani
workers.
The informal money markets were tightened and controlled after September, 11,
2001. Therefore, the inflow of remittances through bank channels increased from
$2.2 billion in 2001-02 to $1.09 billion in 2000-01 (GOP, 2002). The role of
remittances is very much acknowledged in reducing poverty in recent years. The
families and households could maintain or increase expenditure on basic
consumption, housing, education, and small-business formation due to the
increase in the remittances. The total inflow of remittances since 2001-02 to
2005-06 was amounted about $19 billion (Rs.1129 billion). This huge amount of
remittances in Pakistan especially towards the rural areas of the country has
provided support to remove the financial constraints of the receiving households.
They have increased the consumption of durables and non-durables and have
accumulated human capital with education and health facilities. The number of
families has invested their remitted funds in real estate. Pakistan was receiving,
on average, over $ 4.0 billion remittances per annum since from last four years.
This huge amount of remittances provided stability in exchange rate. (GOP,
2005-06).
The micro and macro economic factors have important bearing on the flow of
remittances. The government of Pakistan has made concerted efforts to facilitate
9
the smooth flow of funds. But there were constraining factors which governs the
continuous flow of remittance. The government thrift policies were not
encouraging to attract foreign exchange. The investment climate was not labor
friendly and the investment opportunities were not readily available to the
economic workers. The policy maker can understand a broad spectrum of
migration and remittances in the light of the determinants of the international
migration and remittances in Pakistan. Therefore, the present study will provide
useful parameters for making labour policies and enabling environment for
productive investment in the home country.
1.3 Objectives of the Study
The dynamics of the international remittances were quite comprehensive and
have strategic importance. Due to vast scope of the international remittances,
different researchers have discussed and covered various issues of the
international remittances and these aspects of international migration and
remittances were discussed in literature review. Some of the broad issues
included in the international remittances were; (a) the impact of remittances upon
poverty, consumption pattern of households and, upon investment in the home
country, (b) the trend analysis of the international remittances in which, the
reasons for the fluctuation in the remittances were traced out and explained over
the time, (c) estimation of the determinants of the international remittances, and
(d) some other issues relating to the remittances.
10
However, the objectives of this research study were to explore the major
economic determinants of international remittances in Pakistan. The specific
objectives included in the research study were:
1. Determine the economic determinants of international migration in
Pakistan.
2. Assess the trend of remittances in Pakistan since 1970s.
3. Identify the factors causing fluctuations in the volume of remittances over
different periods.
4. Suggest implications for the policy makers.
1.4 Plan of Study
Nine chapters have been consolidated in this research study. The introduction of
the study has been completed in chapter 1 and the review of literature about the
remittances was given in chapter 2. The theoretical framework of the
international migration and remittances was explained in chapter 3. Chapter 4
was about the data and methodology used in the research study. The unit root
test and co-integration analysis were presented in chapter 5 and in chapter 6
respectively. The determinants of international remittances were given in chapter
7 and the determinants of international migration were given in chapter 8. Finally,
conclusion and policy recommendations of the study were presented in chapter
9.
11
Chapter 2
Literature Review
2.1 Introduction
The review of literature was quite helpful to acquaint with previous related
research work undertaken in the past about the determinant of the international
remittances in Pakistan and different broad aspects of migration and remittances
at national as well as at international level. Therefore, the review of literature
about the estimation of the economic determinants of international remittances
and migration has united different the factors having impact on the economics of
remittances. As a result, the literature survey has provided help to develop the
theoretical framework and to formulate a number of hypotheses that were tested
in the present research study. Testability and reliability of the findings of this
research about the economic determinants of international remittances and
migration in Pakistan provided comparison with other studies elsewhere. The
different results based upon the types of data and econometric techniques used
were drawn by different researchers. Therefore, keeping in view the importance
of literature relevant to research topic, the literature was surveyed about the
analysis of remittances and patterns of migration at national as well as at
international level. The literature about the determinants of international migration
was also reviewed in the research study. The migration was an important
12
determinant of international remittances in the light of empirical findings of a
number of research studies.
2.2 Literature Review at National Level
Amjad (1986) analyzed the uses of remittances in Pakistan by using the survey
data from ILO/ARTEP phase II migration study 1986. The time series data about
total remittances, remittances from Middle East, Gross Domestic Product (GDP)
at factor cost, net income from abroad, per capita income and balance of
payments in Pakistan for the period 1975 to 1985 were used in this study. It was
concluded that remittances financed sufficiently large part of aggregate
consumption, investments and other needs of the economy. It was about 20
percent of total remittances that were invested in the different sectors of the
economy. Pakistan was receiving the remittances from Middle East and these
remittances were at its maximum level in 1982-83. It was estimated that the
remittances from Middle East contributed up to 75 per cent to the overall balance
of payments and debt servicing repayment in Pakistan. The workers’ remittances
from Middle East financed about 36 per cent of the merchandise imports and
non-factor services. The total consumption was increased in the country about
34.6 per cent with the help of remittances.
The increased amount of remittances from the Middle East have also contributed
much in the sectors of manufacturing especially small scale, construction,
13
transport and communication, and wholesale and retail trade. The demand for
basic consumer goods and durables was considerably increased by the
remittances receiving families in the country. Therefore, the remittances have
accelerated the growth of small-scale industries in Pakistan positively.
Adams (1996) analysed the impact of remittances upon income distribution and
assets accumulation in rural areas of Pakistan. Income decomposition technique
was used by dividing remittances into internal and external remittances. Three
years panel data was used from 727 households. Five types of rural assets were
included into asset-accumulation model and these assets were: irrigated land
owned, rain-fed land owned, livestock assets, agricultural capital, and non-farm
assets. It was concluded that rural assets were accumulated in Pakistan due to
remittances. The accumulation of agricultural capital was due to the internal
remittances and the accumulation of land was due to external remittances. It was
found that internal remittances have a positive effect on income distribution, and
the effect of external remittances upon income distribution was negative. Internal
remittances were earned mainly by lower-income groups and represent an
important component of the incomes of households in the bottom income quintile.
Therefore, internal remittances accounted for only a small part of overall income
inequality and it was less than 3 percent. The external remittances were usually
earned by upper-income groups as this group afforded high entry costs to
migration from Pakistan to other parts of the world. The external remittances in
these areas of Pakistan were a source of income inequality and accounted for a
14
12 per cent share of overall income inequality. Therefore, author suggested to
policymakers in Pakistan to take steps to encourage internal migration and to
more carefully consider the consequences of external migration because internal
remittances have a positive effect on rural income distribution as well as on rural
asset accumulation.
Burney (1988) used the time series data for the period from 1970 to 1986 to
investigate the impact of workers’ remittances from the Middle East on the
Pakistan’s Economy. The ordinary least square (OLS) method was used to find
out the determinants of remittances from the Middle East to Pakistan. The
included explanatory variables in the model were: Pakistan’s exchange rate,
crude petroleum price index, gross investment in the Middle East. It was
concluded that the remittances from the Middle East supported to provide foreign
exchange reserve and the availability of remittances has reduced the current
account deficit in Pakistan. These remittances also reduced the external debt
burden, improved debt-servicing ability and decreased the need for additional
foreign loans. The impact of remittances on private consumption at macro level
was also estimated in the research study. For this purpose, private consumption
expenditure, gross domestic product and expected inflation rate for the period
1959-60 to 1973-74 were included in the regression analysis. It was estimated
that coefficient of remittance was increased to 1.45 with inclusion of expected
inflation in the model, showing that private consumption behavior was sensitive
to the expected inflation rate in the country. The propensity to consume out of
15
remittances was estimated to be 0.85 when unofficial channels of remittances
were included in the model. The elasticity of private consumption expenditure
with respect to remittances was found to be 0.05.
Batzlen (2000) determined the impact of foreign remittances on Pakistan
economy. The primary data was collected mainly from the districts of Gujrat and
Jheilum in the research study. The savings and investment functions of the
migrants from Pakistan, who went abroad for employment, were estimated in the
study and ordinary least square (OLS) method was employed to estimate these
functions. The impact of remittances on capital formation and productive
investments creating employment in Pakistan was analysed. It was found out that
remittances were an important source of financing investments, particularly in
agriculture sector of Pakistan. The achievements of skills have a well-built
positive impact on the investment activities undertaken by migrant households.
The worker migrants who have achieved skills during their overseas coursework
were relatively low paid in the home country. The laborer who went abroad to
countries other than Middle East countries, showed a higher propensity to invest
than those who worked in the oil exporting countries. Finally, a relatively large
number of investments undertaken by migrants were characterized by low
productivity expressed in the return on these investment projects.
16
Hyder (2003) established a relationship among workers’ remittances, premium
and resident Foreign Currency Account (FCA) for Pakistan during the period from
July 1993 to December 2001.The Johansen’s model selection and maximum
likelihood co-integration technique was employed to estimate the relationship
among the stated variables. The results suggested that these variables were co-
integrated from July 1993 to April 1998. However, the relationship was not
significant after the period of April 1998. It might be due to the two-tier exchange
rate regime, freezing of FCA and tightening of foreign exchange regulations in
the subsequent periods of nuclear tests in Pakistan. In addition, the causal
relationships were unidirectional and were from premium to remittances and
were from remittances to resident FCA before the period of nuclear tests. The
direction of these relationships was reversed after the nuclear tests. The
premium was the important determinate of the exchange rate, payments and
trade policies in Pakistan and it has also impact upon the amount of remittances
in Pakistan. A desirable high premium might undermine the allocative role of the
exchange rate and was the causes of market segmentation because the workers
motivated to efficient and speedy services for the delivery of the remittances. It
was suggested that banking and fund transferring companies, should improve
efficiency and were argued to reduce the time required for the transfer of
remittance from host country to home country. The delivery system of
remittances from on place to other was also required to be improved. Then banks
will be to compete with the informal sectors used to transfer the funds. For this
17
purpose, the investment in information technology and human capital by the
banking system will be required.
Kazi (1988) used the ordinary least square (OLS) techniques to know the impact
of remittances from Middle East countries upon employment, wages, balance of
payments, exports and prices in Pakistan. The time series data from period of
1975 to 1985 was used in the research study. It was found that the rate of
unemployment in the country was positively related to the education, skill
classification and financial savings. It was estimated that the unemployment rate
of returned migrants with college or university education in urban areas was 33
per cent as compared to unemployment rates of 15 per cent and 16 per cent in
the case of migrants with primary and middle school education. The
unemployment of migrants’ worker with savings of between Rs. 100,000 to Rs.
200,000 was 30 per cent as compared to 17.5 per cent of returnees with a saving
of Rs. 25,000 or less. Similarly, skill professional and technical workers had 40
per cent and 29 per cent high rates of unemployment as compared to unskilled
workers who had a much lower rate of unemployment (18 per cent). It was
further concluded that the shortages in the commodity producing sectors and
higher imports prices were the main determinants of the rate of inflation in
Pakistan during the said period. The rate of inflation was negatively related to the
increase in the amount of remittances. The negative relationship among the
remittances and inflation in the country was due to the fact that the increase
18
inflow of foreign exchange through remittances might allow to imports which lead
to a constraining effect on the price level.
Malik and Sarwar (1993) estimated consumption patterns of households in
Pakistan with and without remittances using the data from the Household Income
and Expenditure Survey 1987-88. Estimations were based on ordinary least
square (OLS) using the SPSS PC+ package for three expenditure groups, which
were, consumption expenditure, durable expenditure and total expenditure. The
hypotheses were formulated to test the differences in consumption patterns of
households were: with and without remittance function having the same slope but
different intercepts; with and without remittance function have the same
intercepts but different slopes; with and without remittance function have different
intercepts and different slopes. It was estimated that the marginal propensity to
consume (MPC) was low for those households who were receiving international
remittances than the households receiving only domestic remittances. The MPC
of both domestic remittances receiving households and the international
remittances receiving households was low as compared to the non-migrant
households in the rural areas of Pakistan. The MPC was found highest in the
urban Punjab and was lowest in rural NWFP for the international migrant
households in Pakistan.
19
2.3 Literature Review at International Level
A large body of literature about the remittances at international level was
available. The literature at international level in this research study was divided
into two broad categories; the review of literature at micro level and at macro
level. Usually, the cross sectional data based on survey sample was used for the
studies of remittances at micro level. Such types of studies were conducted to
know the microeconomic determinants of remittances and analyses the impact of
remittances on the migrants’ household, receiving and sending countries. The
published data of time series and panel were used for the analysis of remittances
at macro level. The macroeconomic determinants of remittances were found in
these studies. Similarly, the literature about the international migration was
reviewed in this study at international level.
2.4 Literature Review about Remittances at Micro Level
Alburo and Abella (1992) examined the impact of workers’ informal remittances
to the Philippines economy. The information from a questionnaire-based survey
of a sample of 600 return migrant workers in the Philippines was used in the
study. It was estimated that informal remittances in the form of foreign exchange
were used to finance trade in the country and the impact of such informal
remittances was as have the formal inflows of the remittances in the country.
There was net effect of the remittance on the domestic economy and not only
20
was on trade in case of funds were remitted in the domestic currency. The
remittances might produce multiplier effects’ depending on the kind of
expenditure in the country. It was concluded that the amount of foreign currency
coming in via merchandise exports was lower than capital flight under valuation
of exports was going on for some time. Similarly, there was evidence of import
under valuation to evade high tariff rates on trade regulations. This import under
valuation was also associated with open or technical smuggling. The contribution
of these remittances to trade transactions suggested an even bigger impact that
the lower inflow of exports, through under valuation exacts a greater burden on
the remittances.
Castaldo and Reilly (2007) investigated the impact of migrant remittances upon
the consumption patterns of Albanian households. The data drawn from the 2002
Albania Living Standards Measurement Survey was used to know the impact of
domestic and international remittances upon the consumption patterns in the
economy. The budget share equations were formulated and estimated for four
different kinds of the commodities. These commodities were food, non-food,
durables and utilities. The controlled variables were also included during the
estimation of the formulated equations. These were about whether or not the
household received remittances from within Albania or abroad. The study found
that the consumption pattern for households in receipt of internal remittances
was not statistically different from those not receiving internal remittances. The
households who received remittances from abroad have less consumption of
21
food and have a higher share on consumer durables as compared to households
not receiving any type of migrant remittances.
The estimated coefficient of the variables capturing the receipt of internal
remittances was not statistically significant in any of the formulated budget share
equations. The estimated effect for the receipt of international remittances was
found to be significant for all kinds of commodities other than non-food. It was
estimated that households receiving no remittances at all have 4.5 per cent lower
budget share on food as compared to the households in receipt of external
remittances. The receipt of external remittances induced an increase of about 25
per cent in the household’s budget share of durables and an approximate 16 per
cent increase in the household’s budget share for the utilities category. It was
found that households receiving external remittances displayed a higher marginal
propensity to consume food items relative to those not receiving any source of
remittance. This result might be explained by the fact that households which
receive external remittances did not classify food as a necessary good anymore
and when they became richer they tended to switch from poor quality to better
quality food types. The remittances were tended to increase a household’s
propensity to consume investment-type goods in the context of Albania, where
households faced severe and frequent cuts in the provision of power and water.
Furthermore, with reference to the category of durables, even items that might be
seen as non-investment goods, such as TV and domestic appliances, may have
multiplier effects within the local economy, as increased demand for these types
22
of goods have created incentives for the establishment of new retail businesses.
Higher amounts of remittances from external sources may be expected to make
a significant difference to the overall welfare of the receiving households. It was
possible that migrant remittances may ultimately exert an impact on the local
economy through enhanced investment in small businesses. It was suggested to
the Albanian government in the light of this research study to capture the
household response to remittances and facilitate them for transferring
remittances and also channeling remittances into the productive uses in the
country. It might only be possible by creating the conditions for a stable
investment environment in the country.
Frank (2001) estimated the impact of the remittances upon the economy by using
the data from the Philippine town of Pozorrubio. It was reported that largest
number of migrants from this village were working in the world. The positive
impact of the remittances was estimated upon housing stock, appliance
consumption, public goods including parks, hospitals and clubs. The remaining
family members of this town have investment in education. Frank (2001) also
mentioned the social and economic cost of migration for this village that has to
pay at individual and collective levels. The author noted the increasing number of
spouses of overseas workers engaging in leisure activities and depriving the
nation of contributions they could otherwise make to productivity growth and to
output. Over time, the people of the village became dependent upon only the
23
amounts of remittances with passage of time. Therefore, the flow of remittances
might hinder economic development of the economy in the long run.
Swamy (1981) analysed the role of remittances in different countries including
Greece, Yugoslavia and Turkey. The socio-economic and demographic factors
were responsible for determining the size of remittances in these countries. It
was concluded that the amount of remittances was determined by the economic
conditions in the host country. The economic growth and progress in the host
country raised the demand for the economic workers and they have more
opportunities for earning in the host countries. Therefore, the level of remittances
goes up due to the increase in economic activity in the host country. It was
further found that 70 to 95 per cent of the variation in the amount of remittances
was due to the fluctuations in the economic conditions of the host countries. A
further analysis in this regard by Swamy (1981) showed that the number of
migrant workers abroad and their wages together explained over 90 per cent of
the variation in the remittances. In the context of the per capita remittances rather
than the total amount of remittances in these countries, time spent abroad by the
worker and number of children were the determinants of remittances. Lengths of
stays abroad, and demographic variables such as number of females’ migrants
were also found to be the determinants of remittances.
24
Benoit and Vencatachellum (2002) estimated the determinants of the internal
remittances in Tunis. The primary data was collected from 500 domestic workers
in Tunis for this purpose. It was the basic characteristics of the sample that all
females were included in the sample and half of them were younger than 18, and
were declared the child domestic workers category. The family of Tobit models
was employed for the estimation of the determinants of remittances. The
remittances were divided into compulsory and voluntary remittances. The
compulsory remittances in the shape of full wages were obtained from the female
workers having age less than 18 years old.
The results of the study showed that the family gender composition has an
asymmetric impact on compulsory remittances, which were levied by the parents,
and voluntary remittances, which were sent by the domestic worker. There were
more compulsory remittances with the increase in the number of young females
in the family because females were remitting funds to their family. The voluntary
remittances were found the increasing function of the number of young brothers
and independent of the number of young sisters. It was very interesting that the
determinants of the decision to remit were different from the amount of actual
remittances. It was concluded that parents have full control over the wages of
their young daughters who were working in different cities. The Income
elasticities of voluntary remittances were high because majority of the domestic
workers included in the survey was young and was also dependent on their
parents.
25
The relative independence of female worker was negatively related to
compulsory remittances but not the actual amount of remittances. An
independent domestic worker kept her wages and voluntarily decided to send
remittances. An asymmetric impact of male and female workers having age more
than 18 years old on compulsory remittances was found in both the standard and
Type II Tobit models while estimating the determinants of remittances. The
compulsory remittances were not found related with the number of young males
in the family. However, the compulsory remittances were increased in the Tobit
model (Standard) because of the number of sisters older than 18 years old.
Rodriguez and Horton (1994) empirically estimated the relationship among the
education and remittances in the Philippine. It was concluded that remittances
were negatively related with the level of the migrant’s education because the
skilled workers might belong to educated and wealthy families and, therefore,
have less incentive of sending remittances to home country. They also stay
longer in the host countries and have desire to unite with their families and
friends in the country of their stay. Therefore, it was concluded that the amount of
remittances decrease as the stay of migrant workers’ increases in the host
country. Therefore, even a positive relation of education with remittances might
even not be the evidence that the brain drain was associated with a greater
amount of remittances in the home country. The direct effect of skills and
education might be positive, but the overall effect due to the longer stay of the
skilled worker in the host country, might be negative.
26
Faini (2006) established the relation between skill of the migrant workers’ and
volume of remittances empirically. For this purpose a simple model with some
key assumptions was developed. The assumptions of the simple model included
that utility of migrant worker’s was a positively related with the consumption of
migrant worker, the amount of remittances and the number of reunited family
members and friends. Family members were then divided into two groups, on the
basis of their closeness to the migrant. Migrants received more utility by sending
more amounts of remittances to their close family members. The impact of a
change in the composition of migration with higher skills and wage was also
incorporated in the model. The increase in wages brought two contradicting
impacts up on the level of remittances. Higher earnings of the worker in the host
were positively related with the level of remittances and this effect was termed as
Wage effect. The higher earnings of the migrant used to reunite him with the
close family members, has a negative effect upon the level of remittances and
such type of effect was known as a reunification effect. Therefore the total effect
composition change with higher wages was still an empirical research. Finally,
Faini (2006) concluded in the case of Europe that skilled and more educated
migrants might not return to their home countries and preferred to stay in the host
country where as low skilled workers have less chances to unite with their
families in the host country.
27
2.5 Literature Review about Remittances at Macro Level
The literature surveyed at macro level to know the determinants of remittances
for home country revealed a number of main factors responsible for the level and
amount of remittances in the home country. The (Russell, 1986) pointed out that
the macroeconomic determinants of remittances emphasized the number of
workers, wage rates, economic activity in host and home countries, exchange
rates relative to interest rate between labor sending and receiving countries,
political risk factors in sending country and, facility of transferring funds. (Sakka
and Mcnabb, 1999) identified that inflation, interest rate differentials and
efficiency of the banking system was the main determinants of remittances.
Wahba (1991) suggested that political stability, consistency in government
policies and financial intermediation significantly affected the flow of remittances.
In a sample of five Mediterranean countries, Faini (1994) found that real
exchange rate was also a significant determinant of remittances. According to
Swamy, (1981) the real earnings of workers and total number of migrants in the
host country were consistently found to have a significant and positive effect on
the flow of remittances. Straubhaar (1986) found the relationship among the
exchange rate and remittances in case of Turkey. The time series data during the
period 1963-82 was used to know the said relationship. The study explained that
government policies for more return on the deposit to increase the flow of
remittances and changes in the exchange rates could not bring more remittances
in the country.
28
Aydas, and Ozcan, 2004 found in case of Turkey that the black market premium,
interest rate differential, inflation rate, growth rates, home and host country
incomes and periods of military regime in the country significantly affected the
flow of remittances. Chami etl, (2005) found a significant negative relationship
between the income gap of the recipient country against the US and worker
remittances in percent of GDP. Swamy, (1981) and other studies found out that
level of economic activity, real earnings of workers and total number of workers
in the host country to have a significant and positive effect on the flow of
remittances. However, the impact of relative rates of return, exchange rate
premium, domestic income and inflation was rather mixed. In view of Glytsos
(1988), neither interest rate differentials between the host and home countries,
nor the variation in exchange rates have any effect on remittance flows. Katselli
and Glytsos (1986) said that per capita remittances were related to the interest
rate in the host country. Chandavarkar (1980) found that realistic exchange rates
and existence of necessary institutional environment significantly affect the level
of remittances.
Buch and Kuckulenz (2004) considered that demographic factors have impact
upon the level of remittances in any country. In these factors, the number of
female employed in host country or a high age-dependency ratio in these
countries were negatively related with the level of remittances. The rate of
illiteracy has impact upon remittances positively. Lucas and Stark (1985) have
detailed discussion about the motives of remittances. According to them, the
29
motives of remittances might be purely altruistic, self-interest, or might be due to
a mutually beneficial agreement between the migrant and the family left behind.
Badawi and Rocha (1992) developed a remittances model incorporating two
approaches in it. The first set of determinants of remittances was related to
worker and family characteristics. According to this approach, the required level
of remittances dominated by income and demographic factors and very less
influenced by policy of government. The second set of determinants was related
with the portfolio approach. According to this approach, the residual level of
remittances was determined from the macroeconomic environment in the home
and host countries.
Sakka and Mcnabb (1999) estimated the macroeconomic determinants of
remittances in Egypt. The data was used from the period of 1967 to 1991.
Ordinary Least Square (OLS) regression technique was employed in their study.
It was estimated that the wage rate of worker, level of domestic income, domestic
price level, the domestic and world interest rates, and the official and black
market exchange rates were the determinants of remittances. The authors
explained the mechanism in which inflation might influence the level of
remittances in the country. The high rate of inflation in the home country reduces
the real level of the income and caused to increase migration and amount of
remittances in the economy. It was noted that there would be less pressure for
migrants to send more remittances in foreign currency in case of depreciation of
30
domestic currency due to inflation. It was also stated that remittances might be
dependent on the allocation decisions for investment projects. The amount of
remittances was determined by the domestic rates of returns. In case of the
interest rates were found low as compared to the host country's, then migrants
will send fewer amounts of remittances. The authors suggested that in case of
black market, the workers will have choice whether to remit through official
channels or other ways in case of black market. It depends upon the difference
between the official and black market exchange rates. The workers will also
adopt unofficial channels in case of taxed remittances by the government. It was
also estimated that high wage rates were positively related to the level of
remittances. The current level of remittances as well as the lagged remittances
was positively related to the domestic GDP. The GDPs of home and host
countries were important determinants of remittances.
Koksal (2006) estimated the micro and macroeconomic determinants of workers’
remittances in Turkey and also analysed the impact of remittances upon the
economy. The Central Bank of Turkey (balances of payments) and Turkish
Statistical Institute (national accounts) were the main sources of data used in the
study. It was concluded that commercial banks in Turkey and the Central Bank of
Turkey were very important for attracting remittances into the country. The major
objective of the establishment of Turkish banks in the field of migrant remittances
was to facilitate and channel the remittances into the country and to achieve the
macroeconomic stability and earning profits from the migrants’ money through
31
transaction fees were the secondary objectives. The analysis of the impact of
remittances upon the economy explored that the impact of remittances on
Turkish economy remained very limited in spite of the best practices for
channeling remittances towards development. The set of variables that explained
the determinants of remittances were the dynamics of family ties and
macroeconomic stability: The family ties included the factors such as the social
status, well-being and risk-sharing by migrants and their relatives where as
factors such as inflation, growth, interest rate differentials and exchange rate
were included in the macroeconomic stability.
Schiopu, and Siegfried (2006) estimated the determinants of migrant workers’
remittances from Western European countries. A sample of neighboring
countries was selected and a country-by-country data set was constructed to
incorporate remittances and non-bank transfers in small amounts. The various
aspects of the remittances were considered in the study because data provided
the information about the flow of bilateral remittances. The data contained the
information about the GDP differences between home and host country, the
difference in returns to financial assets in the two countries and costs of
remittances, the size of the financial network between two countries. The
information was also available about the skill of worker, income inequality and
the share of the informal economy in the sending country. The dummy variables
were also used to capture the impact of different time periods upon remittances.
32
A number of specifications of the model were checked in the study to know the
suitable combinations of explanatory variables.
According to the results of the study, the average remittances were positively
related to the income differential. It meant that remittances on average, people
were more in case of the home country was poorer than the host country. This
result was true in the light of remittances theories. The impact of real interest rate
differential was not significant in this case. In the light of these results, it was
concluded that the remittances were sent to home on basis of altruistic motives
rather than investment purposes. The average amount of remittances was
reduced due to large number of unskilled worker in the country because these
workers have lower wages to send remittances to their countries. Gini coefficient
was used as a measure of income inequality in the host country. Skilled workers
have more opportunities for earning in the countries having unequal distribution
where as unskilled have more in case of equal income distribution. It was
concluded that remittances from countries with a higher share of low skilled
workers was to be lower. This was an important result since existing evidence on
the relationship between remittances and migrants’ skill composition was very
limited. The effect of income inequality in the sending country was inconclusive.
In case of GDP of the sending country included in the model, its impact was
positive, without affecting any of the other results and the cost of remittance did
not affect the amount of money remitted significantly. Finally, if official record was
linked with unofficial remittances, this should depress the total amount of officially
33
recorded remittances. It was found that a larger share of unofficial activity in the
economy lowered the amount of (official) remittances per worker because worker
used informal ways to remit money with the rise of unofficial channels.
Chami et al (2005) developed a mathematical model to know the determinants of
remittance flows. The panel data was used in the study for estimation of the
determinants of the remittances for the period from 1970 to 1998. The data from
World Bank data base was used for 113 countries from all over the world and
both microeconomic and macroeconomic variables were included in the data set.
It was found that remittances were not source of capital of accumulation but
these have negative effect upon the economic growth of the remittances
receiving countries. The basic assumptions of the model used in the study were:
The motivation for remittances was altruism; remittances were compensatory and
counter-cyclical in their nature. It was explored that receipt of remittances
affected the decision to work because the family members who did not migrate,
participated in the domestic labor market and contributed the economic growth
and development in the country. It was also observed that migration brought
moral hazard in the country because workers were so far away from their families
to look after their children. Consequently, remittances might cause a negative
and slow down of economic growth in the country.
34
Jovicic and Mitrovic (2006) explained the determinants and impact of remittances
on development and poverty alleviation in the South East Europe (SEE) by using
econometric techniques. The impact of European Union enlargement policy was
especially tested with the migration and remittances. The dynamics and
prospects of remittances per migrant pointed out the increasing importance of
remittances the over time. It was due to more rapidly rise in the real GDP per
capita and as a result of higher migrants' incomes in the EU host countries, and
reductions in the number of migrants as compare to population in the migrants'
home countries. It was revealed in the comparative analysis of the SEE countries
that differences in relative remittance levels among countries were due to the
income gaps among the different countries, demonstrating that SEE countries
with lower and higher GDP per capita received high rate of remittances per
capita as compared to the middle-income SEE countries.
The most important internal determinant of remittances found in the SEE
countries was domestic labour market situation. The remittances were increased
with increasing level of unemployment at the domestic labour market. It meant
that more jobs in the SEE countries would destabilize the level of remittances,
and therefore migration pressures to decline. GDP of SEE countries were found
an important determinant of the remittance for the low-income recipient countries.
Remittances were found more intensive during the down-cycles of the economic
activity in these low-income SEE countries. The remittances increased relatively
in periods of crises in case of remittances were behaving counter cyclically.
35
Remittances were not impact upon the living standard of migrant families in
higher-than-average-income SEE countries and growth of the home country
might attract more remittances for investment purposes. The impact of GDP per
capita level on the amount of remittance was found positive in case of higher-
income SEE countries.
Rahman (2003) investigated impacts and determinants of foreign worker
remittances in the economy of the Kingdom of Saudi Arabia (KSA). The time
series data for the period from 1975 to 2001 was used in the study. The
stationarity of the time series were seen in the study. The error correction model
was employed to obtain the relationship among the included variables in the
study. The included variables in the study were the real GDP as income
variables, wages per worker, returns and parity conditions, plus some composite
indices pertaining to socioeconomic factors and risk indicators in the Kingdom.
Patterns of the remittances over the time, total and per worker remittances were
explored in the study. The remittances were found to be pro-cyclical with activity
in the Kingdom increasing during booms. The wage rate was a positive
determinant of the level of per worker remittances. A positive relationship
between the level of per capita GDP and workers’ remittances was found from
the Kingdom. The nominal and real interest rates which measure the return in the
country and differential parity conditions in the host and home countries were
negatively related with the remittances. As far as the risk variables were
36
concerned, the results of the study indicated that the stability of government, law
and order situations, affected remittances negatively and increasing risks were
represented by lower index scores and resulted in higher remittance outflows
from the Kingdom. The models using composite socio-political instability indices
were also experimented. The level of remittances was found negatively related to
instability. It meant that higher the level of instability in the KSA, the lower the
composite risk score and the higher will be the flight of remittances from KSA to
other countries. It was also found that per worker remittances were more elastic
with respect to wages as compared to per capita incomes. Remittances were
found less interest inelastic in number of experiments performed.
Schrooten (2005) estimated the determinants of foreign remittances for the
number of socialist countries. The panel data for the period 1990 to 2003 was
used in their study to estimate the determinants of international remittances. The
model was specified in logarithmic form; therefore estimated coefficients were
interpreted as elasticities. It was concluded that remittances were the source of
external finance in these countries. Two models were specified in their research
study to calculate the determinants of the remittances. Remittances per capita
(REM 1) and remittances in percentage of GDP (REM 2) were kept as dependent
variables in these models. The main independent variables used in the study
were: (i) GDP per capita was used a proxy for income level in the home country
to denote the economic conditions in the home country, (ii) Unemployment rate
was used to denote the situation of domestic labor market., (iii) The sum of
37
exports and imports over GDP were used to denote the standard indicator for the
openness of the economy, (iv) The growth rate per capita of the economy was
used to denote future economic situation of the economy. (v) The spread rate
was the difference between lending and deposit rate was used to know the
performance of the banking system in home countries. Moreover, two dummy
variables were used to capture the general institutional situation of the countries.
The results of the both specifications in estimating the determinants of
remittances were found similar and were stable over the time. It was found that
the remittances increased due to the problems in the domestic economy. The
coefficient of the lagged dependent variable was estimated to be 0.5 for
remittances in percentage of GDP and 0.66 for remittances per capita. It meant
that a rise in per capita GDP by 1 percent led to decrease of REM 1 by 0.8
percent. The short term impact of income level was observed the same for REM
2 model where as the long term impact of per capita GDP differed in both
models. A high unemployment rate in the home countries was highlighted as
main determinant of migration. The REM 1 increased by 0.22 and REM 2
increased by 0.29 percent in the short run due to one percent increase in the
unemployment rate in the home country. In the long-run, the one percent
increase in unemployment led to an increase of REM 1 by 1.3 percent and to 0.7
percentage increase of REM 2. It was found that growth has very small impact
upon remittances in the short-run and has no effect in the long run as the
coefficient of growth was zero in long run. The increase in the amount of
38
remittances was associated with the higher transaction costs in the domestic
banking sector.
Jongwanich (2007) examined the impact of workers’ remittances on growth and
poverty in some selected Asia-Pacific countries. The panel data was used for the
period from 1993 to 2003 in the study. The growth and investment equations
were specified and Generalized Method of Moments (GMM) was employed to
estimate these equations. The problem of endogeneity might arise from including
lag independent variables was also controlled in the model. The human capital
and poverty equations in the model were transformed into Instrumental fixed
effects. Results of the study showed that remittances were the cause of the
decrease in poverty due to the increase in income, smoothing consumption and
easing capital constraints of the poor, they have only a marginal impact on
growth operating through domestic investment and human capital development.
The results also showed that there was no direct impact of remittances on
growth. The remittances might have indirect impact on economic growth through
household credit availability. The domestic investment and human capital might
expand through this indirect impact of remittances. The equations including
investment and human capital were estimated in the study to know the indirect
impact of remittances. The remittances were found positively related with human
capital and investment in these estimated equations. One per cent increase in
remittances was related with an increase in human capital by 0.008 per cent and
39
investment by 0.03 per cent. These results explained that remittances might ease
credit constraint and positively affect private investment. The remittances were
used to finance education and health. It meant that human capital was improved
as the impact of remittances was concerned with human capital. The remittances
have created marginal positive impacts on economic growth through the
channels of investment and human capital. One per cent increase in remittances
was related with an increase in economic growth by only 0.03 per cent. It was
noted that other variables in growth equation were statistically significant and
have expected sign. The negative coefficient related with initial income has given
support to the conditional convergence hypothesis. It explained that poor
economies moved to grow faster than rich economies, once the determinants of
their steady state were detained constant. The positive and significant coefficient
of openness pointed out that trade liberalization was useful policy to Asian and
the Pacific countries in promoting economic growth. By contrast, an increase in
inflation and government consumption tended to retard long-term economic
growth.
It was also estimated that an increase in remittances could directly lead to
poverty reduction. Other things considered constant, 10 per cent increase in
remittances has reduced the poverty incidence by 2.8 per cent. It showed that
remittances could directly increase income of poor people, smooth household
consumption and ease capital constraints. In addition to the direct impact,
remittances have also indirect effect on poverty reduction because remittances
40
affect economic growth and human capital, were considered key determinants of
poverty equation. Hence, an increase in remittances of 1 per cent could entirely
improve poverty incidence by 0.43 per cent.
The authors also highlighted the apprehension of remittances with income
inequality. It was because the international migration could be a costly risk so
that the households capable of producing migration and sending remittances will
be comfortable in this context. Poor households would not get the benefit from
the flow of remittance, because of having a tendency to create inequality, poverty
could increase over time. However, the coefficient related with inequality tended
to be less than that of growth and human capital so that the negative impacts
from inequality were improbable to control positive impacts arising from growth
and human capital. This result recommended that remittances could generate
incomes even for families who receive no remittances at all mainly through the
multiplier effects of extended expenditures. As the families increase their
consumption of services or goods produced in sectors with excess capacity due
to remittances, the additional demand could create jobs for other families who in
sequence spend and create further demand. Thus, such multiplier effect could
lead to poverty reduction even some poor families did not directly get remittance
inflows.
41
The remittances should not be considered the key instrument of the growth and
development like exports and foreign direct investment (FDI) in promoting long-
term economic growth in the economy while formulating the development policy
of the country. However, remittances have a significant impact on poverty
reduction; therefore the governments of host and origin countries should aim to
file the impacts of international flows of remittances, considering the poor people
especially. It might be recommended to sharpen the impacts of remittances that
government should have the policy scheme to enhance the amount of
remittances, particularly through formal channel. Secondly, policy scheme should
be emphasized toward the productive use of the remittances. The physical and
human capital investments were found key channels through which remittances
could generate the positive effects on economic development according to the
results of the study. Measures that encourage remittances to such investments
would enhance its developmental impact. They could be undertaken in various
forms. For example, government could develop appropriate training and
educational programs in the country for remittance receipts to make effective
investment decision. In addition, the appropriate infrastructure should be
developed to generate favorable investment climate.
Puri and Ritzema (1999) reviewed different systems that seek to channel
unrecorded remittances through formal banking channels in many Asian and
Arab countries. They attempted to separate the main features of recorded and
unrecorded remittances. This method to watch the impact of remittances on the
42
domestic economies of labour-sending countries concentrates completely upon
the officially recorded remittances and their effects on the different
macroeconomic variables in the formal economy. It was recognized that officially
recorded remittances usually fall short of actual overseas savings of migrant
workers and that the difference was of particular significance in countries which
have trade and exchange-rate restrictions and unsteady domestic economies,
the true magnitude of unrecorded remittances and their economic implication
have received relatively less importance. There was subjective evidence that
only a fraction of the unrecorded remittances represent pure leakages and that
the largest part of remittances through informal channels to finance domestic
consumption, investment and foreign trade dealings in the migrants’ country.
Therefore, unrecorded remittances have become one of the most critical
dimensions of the remittance systems in many Asian and Arab economies.
The authors also looked at policy measures adopted to influence and optimize
the use of remittance in the domestic economy of these countries. The potential
role of microfinance with regard to the scope for linking unrecorded remittances
and investment was evaluated. A major finding of their study was that remittance
leakages were, to a large extent, an indication of the macroeconomic policy
regimes of labour-sending countries. Therefore, the best solution to the problem
of increasing their developmental significance would be to implement wide
ranging policy reforms aimed at setting the macroeconomic variables in order to
encourage remittance inflows through official channels by using micro-finance
43
tools and improving the existing banking network to effectively compete with
informal market arrangements so as to channel the funds into productive
investment.
Lu and Treiman (2007) calculated the effect of remittances sent home by South
African Black labor migrants on children’s schooling. They used cross-sectional
data for the period of 1993-1994 from Integrated Household Survey and panel
data for the period of 2002 and 2003 from South African Labor Force Survey. It
was found that both labor migration and the likelihood of sending remittances
home were much common among the Blacks as compare to other racial groups.
Therefore, the study was limited to the impact of migration and remittances upon
children’s education in case of Blacks. The receipt of remittances significantly
increased the possibility that children were in school, through three pathways:
increased household educational spending, reduced child labor, and mitigation of
the negative effect of parental absence due to out-migration. Children in
households without remittances were not privileged as compared to beneficiary
households of remittances, and in some respect were even worse-off than their
counterparts in non-migrant households, primarily due to the harmful effect of
parental out-migration with no economic compensation.
The sensitivity tests using fixed-effect and random-effect model showed that the
effect of labor migration and remittances were robust to unobserved
heterogeneity and relatively consistent across sub samples and independent
44
samples over time, although the negative effect of living in households with out-
migrants but no remittances was substantially by 2002-2003, due at least in part
to relaxed migration policies after the breakdown of apartheid. The social
consequences of remittances were also assessed in the study. It was found that
remittances helped to reduce intra-familial gender inequalities as well as inter-
familial inequalities in schooling.
2.6 Literature Review about Migration
Walsh (1974) estimated the determinants of migration in the Ireland. The time
series data was used and included variables were: net migration flow, wage and
unemployment. The net migration flow from Ireland was the dependent variable.
The wage and unemployment were the explanatory variables. The data for the
period 1951 to 1971 about the explanatory variables was obtained from Ireland
and Britain. The results explained that Irish net migration was responsive to
relative labor market conditions in Ireland and Britain. The unemployment and
wage rates differentials among the home and host countries were found
significant determinants of the net migration from Ireland.
Mayda (2007) estimated the determinants of migration inflows into fourteen
OECD countries by country of origin, for the period of 1980 and 1995. The effect
of average income and income dispersion in host and origin countries on
migration was analyzed. The impact of geographical, cultural, and demographic
45
factors as well as the role played by changes in the host countries’ migration
policies was highlighted in their study. The study explained the determinants of
international migration flows and provided a framework for the impact analysis of
migration with time-series and cross-country variation in an annual panel data
set. The study looked at the determinants of international migration, and provided
a framework to analyze the impact of migration in home and host economies
including the impact on living standards.
It was found that the improvements in the mean income opportunities in the host
country (“pulls factors”) significantly increased the size of emigration rates. The
impact of declining levels of per worker GDP (“push factors”) in the origin country
was often negative. The “push factors” suggested that migration quotas were
more obligatory than “pull factors”. A possible explanation of the asymmetry
between push and pull factors was provided in their study. The distance was
found one of the most important factors among the variables affecting the costs
of migration. Its effect was negative, significant and quite steady across
specifications. The demography, in particular the share of the origin country’s
population who was young, was a significant determinant of emigration rates.
The author found that the effect of both variables was more pronounced in those
years when a host country’s immigration laws became less restrictive. The author
also explained the importance of the demand side of the host countries’ migration
policies. In the theoretical framework of the model, it was assumed that migration
quotas were exogenous. Therefore, the endogenous determinants of migration
46
policy were not investigated. The results recommended that migration quotas
matter a lot. The quotas mitigated supply-side effects causing pull and push
factors, independent geography and demography. The consistent evidence with
the constraining role of migration policies was found in the study. It was found
that pull effects became more positive and push effects turned negative in those
years when a host country’s immigration laws became less restrictive.
Braucker and Schraoder (2005) proposed a migration model with heterogeneous
agents and persistent cross country income differentials that features temporary
migration and have examined the macro determinants of migration in theoretical
as well as in empirical perspective. The sample employed in the study was
comprised the migration data from 18 European source countries to Germany for
the period of 1967-2001: the 14 other Member States of the 'old' EU, Iceland,
Norway, Switzerland and Turkey. This sample covers all European source
countries with the exception of the countries of the former COMECON, Albania,
and the successor states of the former Yugoslavia. The data on migration stocks
and flows was used from the German Federal Statistical Office.
The model in the theoretical part established a long-run equilibrium, in which
individuals can stay their entire life in the home country, migrate temporarily
abroad or stay permanently in a foreign country depending upon their choices.
The migration stocks and flows were generated in the model. It was found that
the number of migrants, the duration of migrant stay and the stock of migrants all
47
increases with the income differential between the host and the home country,
while net migration ceases to zero. The gross emigration and return migration
rates were related to the stocks of permanent and temporary migrants. Existing
empirical migration models, estimating net migration flows, instead of stocks,
may be mis-specified because the stock of permanent and temporary migrants
were a positive function of the income differential. The determinants of
international migration were estimated in the empirical analysis.
The results of the panel unit-root and panel co-integration test showed that the
standard flow migration model was mis-specified for the data set used in the
study. The traditional migration model in the empirical model explained migration
flows by a number of explanatory variables including GDP per capita,
unemployment rates, lagged migration stocks and institutional variables. The
macroeconomic variables such as GDP and employment were non-stationary
variables having order of integration one. The tests carried out in the empirical
part of the study indicated that migration rates were stationary, while migration
stocks were I (1) variables. The empirical analysis suggested that the hypothesis
of a co-integration relationship between migration stocks and the explanatory
variables cannot be rejected for the data set used. This was interpreted as
empirical support for the theoretical hypothesis that migration stocks and
explanatory variables such as the income differential and employment variables
form an equilibrium relationship.
48
The results of the study might be used for policy implications. The flow model
explained that migration did not stop before expected income levels among host
and home countries have converged to a certain threshold level, which was
determined by the costs of migration. In case of determined differences in
expected income levels, either the total population will eventually migrate or
migration will not occur in the first place. In contrast, the stock model predicted
that migration ceased when the benefits of migration equal the costs to the
marginal migrant, such that a long-run equilibrium between migration stocks and
expected income emerged. Consequently, migration might stop despite the
existence of large income differential among the countries.
Poveda (2007) analyzed the determinants of rural population migration in south
of Veracruz state (Mexico) by listing three different regions of migration.
Traditional markets, the northern border and the United States were included in
the said regions. A three-level multinomial logistic model was applied keeping in
view the individual, family, and local characteristics of the migrants. The different
determinants of migration were found in different cases. The difference in the
determinants of the migration was due to the differences in the objectives, needs
and income of the migrants and their households.
The results of the study showed that the determinants of migration to the
traditional markets were related with certain personal and family characteristics
and with local conditions. A larger risk of migration in agricultural areas of thickly
populated villages was found where families with a lesser area of agricultural
49
land. These areas were connected to the urban markets in the region and
preferential contact and migrations. Some of migration were supported by family
members and have attracted recent migrants from these areas. The continuation
of migratory previous circumstances was only a significant family determinant in
migration to traditional markets and without background the risk decreased to
35.2 per cent with respect to antecedents. The sex, age and marital status of the
migrant were found the personal determinants affecting the probability of
migration. The young and single man has more probability of migration to the
nearby places.
The home variables played an exceptional role in the process of migration to the
border areas. The growth rate of population was a one of the factor that affected
the risk of migration. The areas of greater population growth rates were related
with greater migration to the border during the period of 1996-1999. This relation
led to definite settlement and hence, the lack of local migratory background in the
region. The saturation of the resources in the area have led people to migrate to
the northern border instead of within the region due to the difficulty in finding a
job in the region and secondly, the greater earnings offered by the cross-border
assembly plants. The place of residence has also affected the migration to the
border areas.
The migration to the border came slow down from the areas where there was a
greater availability of land. The border as well as cross-border, there were job
50
opportunities for the worker due to assembly plants. The migrants used their
income earned from these areas to improve their living conditions. They saved
very less for the investment motives to improve the production conditions. The
migratory antecedents was found significant variable of migration in these areas
due to family members or the migrant himself has experience of migration toward
the border. The specialization to work in assembly plants needed a little skill from
the workers. Hence, education of the worker was the main determinants of
migration to these areas and the higher the educational level was required for
better work prospects.
Davis and Stecklov (2002) explored the impact of rural migration upon the
Mexican economy by employing the data from the rural areas of Mexico. The
data from landed households in Mexico was used to avoid the ambiguous
consequences. Different hypotheses about the choice for the host country were
tested in their study. Hypothesis was maintained that people from rural areas of
Mexico could migrate within Mexico to seek jobs in the agricultural and other
sectors as well as to the United States. The composition of migration was
consisted from rural to rural, rural to urban, and rural to international migration in
their analysis. A simple aggregated migrant network was used as an explanatory
variable in the used approach.
The results of the research showed that the characteristics of migrants to the
USA and non-agricultural Mexico were similar. The level of education of Mexican
51
agricultural migrants was found low, they were indigenous, and their living style
was isolated. The USA migrant networks appeared more significant as compared
to Mexico migrant networks in respect of influencing the migration decision to the
relevant countries. Once networks were disaggregated by kinship, Mexico
migrant networks became very important to the Mexico migrant decision. Failing
to disaggregate migrant networks by kinship relationships and migration
experience were led to inaccurate results.
The impact of migrant networks in the decision to migrate was not homogeneous,
but was depended upon the composition of the network. In particular, the closer
the kinship bond, the more important the impact was found. The impact of
migrant networks was found to be non-linear, but may be increasing or
decreasing at the margin depending on the type of asset and destination choice.
Important interaction effects were found not only among different types of USA
and Mexico assets, but were also between USA and Mexico migrant networks.
The USA and Mexico assets served as substitutes in terms of USA migration,
and complements for Mexico migration. Finally, in confirming the importance of
networks, the results showed that the location of network migrants within the
migrant destinations affect the location decision of subsequent migrants.
Garip (2006) formulated a model to capture the migration decision of individuals
and remittance behavior and found that remittances and migration were
interrelated actions. The hypotheses on the determinants of migration and
52
remittance were combined with a single framework. The data of internal Thai
migration was taken to test this hypothesis. The results of the study showed that
a common set of economic and social factors influenced both migration and
remittance behavior. It was found that rural villagers in Thailand were more likely
to migrate and send remittances in case of their households were relatively
deprived of land or other assets with respect to other households in the village.
The altruism was the main motivation for young adults to migrate to urban
centers and send remittances to support their families in the home. The social
factors were also found to be very important for migration and remittance.
Specifically, individuals were more likely to migrate when there were prior
migrants in their household or community, which suggested the importance of
social networks. Reciprocity obligations within the household and remittance
norms within the village also seemed to have important effects on migrants’
decisions to remit.
The results of the study further explained a common set of social and economic
factors that determine the pattern of migration and remittance. Specifically,
economic needs of the family, the village networks and norms for facilitating the
migration and remittances were the determinants of migration and remittance.
Methodologically, the findings showed the need for jointly modeling migration and
remittance behavior, while taking into account potential endogeneity and sample
selection biases.
53
Demet and Tansel (2007) presented research findings on the return intentions of
Turkish professionals residing abroad. This study used a descriptive framework
to establish the validity of several proposed models of non-return. The results
were based on an internet survey of Turkish professionals abroad.
Correspondence analysis was used to examine the relationship between return
intentions and various factors that may affect this intention. The results
emphasized the importance of student non-return versus traditional brain and
appeared to complement the various theories of student non-return. The
respondents appeared to come from relatively well-to-do families with highly
educated parents. Many have earned their degrees from universities that have
foreign language instruction. The recent economic crises in Turkey have
negatively affected return intentions. It was verified that return intentions were
indeed linked closely with initial return plans, and this relationship weakened with
stay duration. Specialized study and work experience in the host country was
also appeared to explaining the incidence of non-return. Return intentions were
weaker for those working in an academic environment.
These results led to important policy implications, some of which include the
training of individuals for academic positions at domestic institutions, supporting
study abroad for shorter periods and improving academic facilities in Turkey’s
newly established universities. The government may support public and private
Research and Development (R&D) centers to increase the employability of
returnees, but also to improve the quality of the higher education system in order
54
to both reduce the need for education abroad and to increase the attractiveness
of universities as prospective employment places for those acquiring education
and experience abroad.
2.7 Summary
It was concluded from literature review that there was a paucity of literature about
the determinants of remittances in Pakistan. Majority of studies at national level
used cross sectional data to determine the impact of remittances upon the
economy. However, time series data was also used to find the determinants of
remittances in Pakistan without applying latest techniques. These studies have
used ordinary least square method to estimate the determinants of remittances.
The results of such studied were not reliable because the OLS method produced
spurious results due to non-stationary time series used in a regression model.
Spurious results indicated a significant relationship among the variables when
there was no such relationship among the variables (W.E., Griffiths, 2001).
Limited studies in Pakistan have used latest techniques for the estimation of the
determinants of remittances as well as migration including Hyder (2003), which
employed Johansen’s model selection and maximum likelihood co-integration
technique to analyze the relationship among workers’ remittances, kerb premium
and resident FCAs for Pakistan during July 1993 to December 2001.
Literature at international level for the analysis of remittances was available in
abundance. These studies estimated the economic determinants of remittances
55
and migration including the microeconomic determinants as well as
macroeconomic determinants using the latest techniques of estimation of the
time series and panel data. Therefore, macroeconomic determinants of
international remittances and migration in Pakistan were estimated in this
research study using co-integration and error correction method. It was
concluded from the literature survey that remittances was influenced by the
following factors: (i) The number of migrant workers at abroad, (ii) The wage
rates at home and host countries, (iii) The economic activity in the host country
and in the sending country, (iv) The exchange rates between home and host
countries, (v) The relative interest rate between the labour sending and receiving
countries, (vi) The inflation rate at home country, (vii) The political risk at home
and host countries, (viii) The facility for transferring funds from host to home
country, (ix) The marital status of the migrant, (x) The level of education of the
migrant, (xi) The migrant whether accompanied or not by dependents, (xii) The
years since out migration and, (xiii) Household income level receiving the
remittances.
It was further concluded that the process of migration mainly depends upon the
factors including: (i) The income inequalities in the home country (ii) The wage
rates at home and host countries, (iii) The economic activity in the host country
and in the sending country, (iv) The migration policies of the home and host
countries, (v) The differential of unemployment rate between the home and host
countries, (vi) The inflation rate at home country, (vii) The political risk at home
56
and host countries, (viii) The distance between home and host countries, (ix) The
marital status of the migrant, (x) The level of education of the migrant.
57
Chapter 3
Theoretical Framework
3.1 Introduction
The research work about estimating the economic determinants of international
remittances in Pakistan was based upon the important theories of migration and
remittances in the literature. The theoretical framework was logically developed,
described, and it elaborated network of relationships among the variables that
were included in the econometric model through interviews, observations, and
literature survey. These variables were considered quite relevant to the given
research topic, Economic determinants of international remittances in Pakistan.
Thus, a good theoretical framework identifies and labels the important variables
in the situation that were relevant to the problem identified. The elaborations in
the theoretical framework thus addressed the issues of why or how certain
relationships existed, and the nature and the direction of the relationships among
the variables of interest (Sekaran, 1992).
This chapter was mainly devoted to explore the prominent pioneer work in the
field of remittances. The useful theories were developed in the history to carry
out the research work on the different aspects of the remittances. The
researchers seek help from these theories for methods and techniques usually
used for model building. The model building was considered very important for
58
the evaluation of the specific research problem. Model specification for
estimating the economic determinants of international remittances in Pakistan
was not an easiest topic in the applied economics. At the end, Hypotheses were
developed for the determinants of the remittances in Pakistan after reviewing the
theories of the remittances and empirical evidences from the research literature
in the light of these theories. The theory of remittances was presented here in
this chapter with brief account of migration theory.
3.2 Theories of Migration
Number of models and theories were available in the economic literature that
explained the reasons and factors of migration. These models were the integral
part of the theory of remittances. The process of remittances was linked with
migration. The research on international migration has focused on explaining why
people move from one area to another. There was no single theory to explain the
causes of international migration but one of the most important economic causes
was obtaining a higher income in the host country, explained the process of
migration. There were also other factors that affect the decision to migrate from
one country to another country, such as family and friendship networks. The
most important studies were: Stark (1991); Massey et al, (1993); Zimmerman
and Bauer, (1995), yet other researchers attempted to explain why international
immigration occurred. The key objective was not to discuss the theories of
59
migration in this research study. Some important theories of migration were
discussed below:
3.2.1 Neo-Classical Theory of Migration
The Neo-classical theory of migration explained the reasons for migration and
these reasons might be economic reasons. According to this theory, the
individual migrant tried to maximize his utility level and this level was determined
from his choice to migrate. In this argument, the government of the home country
is advised to allow the people for migration so that market forces could achieve
the desired level of equilibrium in the economy (Borjas and Miller, 2003). The
microeconomic and macroeconomic variables explained the process of migration
from one country to other country according to this theory of migration.
An individual to migrate at microeconomic level decision based upon cost benefit
analysis (Sjaastad 1962, Todaro 1976). The analysis declared the decision for
migration if net result of cost benefit was positive. The wage differences and
rates of employment among the home and host countries were important factors
for emigration. The improvement in the expected earnings in the host country, in
the form of increased real salary and more opportunities for employment,
increased the probability level to migrate; an improvement in earnings expected
in the country of origin reduced the likelihood of emigration; and a reduction in
the costs (economic and social) of transferring from one country to another
60
increased the emigration under this approach. The costs of migration, distance
between the countries, difference of language and the existence of ethnic
networks in the host country were found to be important determinants in
explaining the process of migration. Government policies have an impact upon
migration decision that brought changes in the number of factors of migration, for
example, developmental policies of government those increase the wage level in
the home countries and reduce the probability of finding jobs in the host
countries. Similar, the human capital approach explained the incentives to
emigrate decreased with age and considered the decision to emigrate as an
investment for which the current costs will be recovered in the future.
In the context of macroeconomic, the neo-classical economic theory highlighted
that real wage differences among different countries was main cause of migration
and flow of capital (Lewis 1954, Todaro 1976). The new international equilibrium
was created at which the wages in real term in all the countries restored at the
same level. The geographic differences in the supply and demand of labour in
the home and host countries were the main factors that guide the decision of
individuals to emigrate. Hence, emigration occurred from countries with low wage
levels to those with higher wage levels. The second was a capital flow from
countries offering high wage rate to the low rate countries. The flow of capital
consisted upon industrial labour intensive capital and accompanies by high-
skilled labour migration. Therefore, net international labour migration was a
secular phenomenon.
61
3.2.2 Push-Pull Theory of Migration
The theory of pull and push factors examined and explained the factors of
migration in both countries including home and host countries. The factors of
migration in the home country determined the choice of migration and proved
helpful in the process of migration. The demographic factors including growth in
population, economic factors like high unemployment and low living standards in
the home country and socio political factors were included in these factors.
These factors were identified as push-factors. The factors of migration in the
receiving country included but not limited to: labor demand, better economic and
political conditions, existing rules and regulation and incentives for immigration.
These factors were identified as pull factors. Any types of migration such as
domestic or international etc. might be examined in the light of pull and push
factors. The negative characteristics at home country were included in the push
factors and the positive characteristics at the center of destination were included
in the pull factors (Datta, 2002).
3.2.3 Dual Labor Market Theory of Migration
In the dual labor market theory of migration, the institutional factors to the Neo-
classical approach were added. Piore (1979) revealed that this theory of
migration highlighted the importance of race and gender in the process of
migration. This theory also explained the labor market segmentation. The labour
62
market was divided into primary and secondary parts according the dual labour
market theory. The primary segment of labour market was characterized by
capital-intensive techniques of production and labour-intensive production took
place in the secondary segment. Skilled workers in the primary segment have
more social status because they were trained to work with advanced capital.
They have higher income and better employment conditions than unskilled
workers in the secondary segment. Jobs at the bottom of the labour market were
almost found in the secondary segment.
Piore (1979) provided three explanations for the demand for international
workers in the modern industrial societies: general labour shortages, it required
to fill the base positions in the job hierarchy, and labour shortages in the
secondary segment of a dual labour market. The last clarification was also
covered by the first two explanations. General labour shortages led to vacancies
at the bottom positions in the job hierarchy. In addition to general labour
shortages, there were specific shortages at the bottom of the job hierarchy
arising from motivational problems. It brought demographic and social changes in
modern industrial societies. Motivational problems came about because jobs at
the bottom of the hierarchy were often related with low social status and because
the opportunities for upward mobility were low. Demographic and social changes
including the decline in birth rates and educational expansion in modern societies
were the causes of the inflow of teenagers. These young workers were willing to
take jobs at the bottom of the hierarchy to attain experience and earn money. As
63
a result of labour shortages at the bottom of the job hierarchy, employers were
compelled to recruit foreign workers.
3.2.4 Migrations Systems Theory
The migrations systems theory included a variety of discipline and analysed the
process of migration. The base of this theory was to synthesis the migration
movements with the relations of macro and micro structures. The whole economy
of the world, associations among states, and forces at regional, national and
international level were included in the macro structure. The social relations
among the workers in home and host countries were included in the micro
structures (Miller and Castles, 2003). Migration system theory was paying
attention on the structure of the world market, especially on the influence
exercised by capitalist relations on non-capitalist peripheral societies through the
actions of the multinationals, governments, etc. International migration occurred
because labour, land and raw materials fall under the influence of market control.
International migration was not so much affected by differentials in income or
employment as by policies towards external investment and towards the
international flows of goods and capital. According to this theory international
migration was a consequence of globalization and dissemination into markets,
since modern capitalism has generated a mass of mobile work force in search of
better opportunities.
64
3.2.5 The New Economics of Labor Migration
This theory dealt with family and family was considered as a single unit in the
light of this theory. This single unit of family was used in the analysis for
migration. The individual migrant worker was considered a subset of the family.
The costs and benefits of the migration decision were shared with migrant and
his whole family. The individual migrant was part of the beneficial contract of the
family members (Bloom and Stark, 1985). Families were benefited from the
income generated from different sources. This phenomenon became a form of
coinsurance. This theory did not reduce the importance of individual activity in
decision-making for migration. The actions and performances of individuals could
be explained in the framework of decision-making unit with his whole family
(Stark, 1991). This theory has established a unique relation with analytical
approach of migration from an economic perspective and the more sociological
view in which human behavior has been examined. Therefore, remittances
among families were integral to migration according to the new economics of
labour migration.
3.2.6 Trans-nationalism Theory of Migration
Trans-nationalism theory of migration was viewed in the light of expanding
globalization, rapid communication system and modern technologies adopted by
the countries. The migrants, due to availability of these modern facilities were
able to have connections with their family members due to globalization and
65
established cross-border activities with their relatives. Trans-nationalism activities
were those activities that were performed on a repeated level at borders (Portes,
1999). The author described trans-nationalism in the light of above discussion as
aimed at power share by interested groups of the society, state governments,
and also trans-nationalism from below to originate from the people and
perceptions of the society. Therefore, it was a useful differentiation that
determines the nature of transnational migration.
Shiller et al, (1993) stated that trans-nationalism entailed methods in which
immigrants form and maintain dimensional social associations that connect with
their home as well as host countries. The stress given in this case was on the
immigrants framing ties and their use of framework to contain and oppose the
difficult situation of their migration occurrence.
3.2.7 Roy’s Theory of Migration
Roy’s theory (1951) explained causes and impact of income inequalities in the
receiving and sending countries upon the decision of migration. The individuals
situated in the upper part of income distribution have less incentive to migrate in
case of home country has greater inequality than the host country and people
situated in the lower part of this distribution have greater tendency for migration.
In other words if home country has less inequality than the host country, the
individuals in the lower part of distribution have less incentive for migration as
66
compare to the individuals at the top of this distribution. The negative relationship
might be established between relative inequality and migration rate in case of
sending country has a very unequal relative distribution of income. This
relationship of relative inequality and migration rate became positive in case of
sending country has a relative egalitarian distribution. Therefore, an inverted U
relation between relative inequality and migration has been established
according to this theory. For example, if average income of home country was
below the average of the host countries and the Gini ratio of coefficients was
close to one (similar levels of inequality in both countries) then the emigration
ratio would be the highest, since everyone would have an incentive to emigrate
due to income differential between the countries.
The Roy’s theory of migration also explained the relationship between
development and migration. The wages were found low and the emigration rate
was also low due to low wages in the start of the industrialization process. The
immigration rates have been increased due to developments in the advanced
countries in long term process. A decrease in the immigration rate was observed
due to reduction of the wage differential among the countries over time (Rotte
and Vogler, 1998). Therefore, an inverted U shape type relationship between
development and migration has been established. The growth of emigrations in
the first episode of industrialization may be examined by the effects of the
demographic evolution occurred with industrialization. It was also due to the
effects of expansion of the networks among emigrated workers and wanted to be
67
emigrated. Moreover, poverty was an impediment to migration because financing
for migration much difficult in these poorer countries (Rotte and Vogler, 1998).
The focal point of migration abroad was a higher income abroad as a principal
cause of decisions to emigrate. The other variables also exert an important
influence on decisions to migrate, including factors of war, racial prejudice and
political stability at sending country. The preference of host destination was also
influenced by the continuation of a network of family and relatives who have
migrated previously to that country (Solimano, 2003).
More scientifically, the amount and direction of flows of migration were subjective
to the following reasons of long run and cyclical in their nature. The author further
highlighted the economic determinants of the international migration theoretically.
These determinants were (Solimano, 2003):
(a) Per capita income or real wage differentials between sending and
receiving countries: immigration flows were positively related to the ratio of real
income in the host country and that of the recipient country. The expected wage
in the place of destination compared with that of the home country was important
while taking decision to migrate. Moreover, in a dynamic perspective, the current
value of expected relative wage streams would be the relevant variable.
68
(b) The business and economic conditions in both sending and receiving
countries: Growing economic development and labor shortages in the host
countries tended to increase the probability of immigrants to find a job. In
contrast, in periods of sluggish growth and higher unemployment, this probability
was lower. While the decision to emigrate depended largely on real income
differentials between countries, the timing of migration seems to be related with
the state of the business cycle in both sending and receiving countries.
(c) Network effects: Empirical analysis of migration flow showed that high
importance to friends and family was attached for the choice of host country
because relatives and friends were more helpful for the support of migrant at time
of migration. The information relevant to jobs and economic conditions of the
host country might be obtained by the close relative or friends living in these
countries.
(d) Immigration policies: Policies in host countries that were unfavorable to
immigration deter migrants, although not completely, as there still remained the
possibility of illegal migration to some countries.
(e) Costs of migration: There were several costs in the process of migration.
These costs might include travel costs and living expenditure in the destination
country and also the cost of searching for a job in the new country. Low level of
income and having less skill migrants were always suffered by such costs, which
69
may in practice be an important inhibiting factor on the international migration of
the poor.
(f) Cultural differences between countries: The patterns of migrations were
affected by the culture prevailing among the sending and receiving countries.
The language, way of living, customs and traditions were responsible for
migration patterns. The migration pattern might be different among individuals
and countries because the cultural habits were founds always different in the
different regions and countries.
(g) Geographical boundaries and distance: Generally speaking, migration to
neighboring countries was tended to be more common as compared to the
countries situated far away. Geography therefore played an important role in
determining the size and direction of migration flows.
3.3 Theory of Remittance
The research work and empirical findings on the determinants of international
remittances in Pakistan was limited both in lack of theoretical understanding of
the remittances process and paucity of data relevant to remittances. Different
theories explained different results of the remittances depending upon the
household welfare or utility functions with the remittances. The net changes in
income of households, including public transfers, also affected remittances, led to
70
provide different results. The amount and structure of the remittances was also
belonging to the areas of the migrant workers whether he or she belonged to the
rural communities or from the urban areas. The number of household members,
the gender structure of the family and sex of the head of family were also bearing
factors for the amount and use of the remittances in the family. It was observed
that some family members migrate; others remain in the place of origin. The New
Economics of labour Migration (Stark and Bloom, 1985) explained that
imperfections in rural credit and risk markets including farmers' inability to obtain
credits, insure against income loss created incentives to participate in migration
by sending family members to work in the city or abroad. According to this view,
migrants played the role of financial intermediaries, substituting for the missing
rural bank or insurance institution. When the migrant workers reached and were
established at their working places in the host countries, migrants provide the
family members at the origin with needed capital, through remittances. Support of
family members was a deeply ingrained in their cultural and habits in developing
countries like Pakistan. The element of support to family members at home
country become stronger in case of migrant worker has wife and children left with
his parents to look after them. Traditional classical or neoclassical models of
migration behavior (Todaro, 1969) did not explain why migrants share their
earnings with their place of origin. However, remittances were a cornerstone of
the NELM, representing one of the most important mechanisms through which
determinants and consequences of migration were connected.
71
3.4 Motives of Remittances
Solimano, (2003) has defined and explain four motives of remittances. These
were: (i) Motive of Altruistic, (ii) Motive of Self-Interest, (iii) Motive of Loan
Repayment and (iv) Motive of Co-Insurance.
3.4.1 Motive of Altruistic
This school of thought concerned that remittances were a commitment to the
household. Remittances were sent to family household due to act of love and
liability of the migrant worker. It has been explained in the remittances theory that
the important reason for migration other countries was poverty. Keeping in view
the motive of altruistic, providing remittances to the family members for their
welfare gives a utility to the migrant.
3.4.2 Motive of Self- Interest
Economic and financial self-interest was main motives for sending remittances to
the home in the light of this motive. It was argue that that migrant worker try to
save more and more at every point in time. Then, the migrant worker took
decision about the type of assets to be bought and the country in which the
wealth to be accumulated. It was obvious that the home country was considered
the best place to invest in the purchase of property, land and assets including
72
financial assets. Such types of investment in the host country might earn higher
profit with high risk attached. The investment might not be permitted in the host
country as well. Therefore, the family of the migrant worker could manage and
administer the accumulated assets of the worker during the emigration period.
3.4.3 Motive of Loan Repayment
It was discussed and explained in the New Economics of Labor Migration
(NELM) theory that in the process of remittances the family was considered a
unit of analysis rather than the individual it self. The families developed an
implicit contract among the family for members to choose to migrate abroad or
stay at home. This contract has dynamic prospects because this contract migt
last for various years or decades. The elements of investment and loan
repayment were combined in the contract. According to this theory of
remittances, the family invested in the training of the out going worker and
financed the costs of migration including the cost on travel and subsistence costs
in the host country. Therefore, migrant worker has obliged to send remittances to
the family members stayed at home. This was referred as the loan or investment
part of the loan repayment motive. The repayment part of this motive started
when the worker settled in the host country and his salary was started to rise with
the passage of time and the migrant started to repay the loan including principal
amount and interests to his family in the shape of foreign remittances. This
showed that the families try to make investment in the asset having higher return.
73
The size of remittances also depended upon other conditions as the income
profile of the migrant (Solimano, 2003).
3.4.4 Motive of Co-Insurance
An alternate of remittances theory as an inherent family agreement among
worker and family members stayed at home covered risk diversification. It was
suggested that for the family as unit to send some family members to abroad for
the diversification of the risk. The migrant supported his family members in times
of need at home. The remittances played the role of an insurance claim in this
motive. However, in principle, such types of contracts were implicit because
these were within the family members, based upon faith within the family
members (Solimano, 2003).
Keeping in view, the theories of migrations and remittances and motives of
remittances, the economic determinants of remittances could be estimated with a
suitable modeling strategy. Stark and Bloom (1985) considered family as an
appropriate unit for the analysis of migration and remittance, because all the
family members shared and traded off the expenditure and payback of the
remittances. In this way, the theories of remittances explaining the different
dimensions of the remittances concentrated upon roles of the family and
relationships of family members. The remittance choices could be determined by
the role played from such relationships among the family members. Majority of
74
economists before the advent of the theory of New Economics of Labor Migration
considered family relationships as a form of mutual compassionate. This form of
cooperation among the family members was a prime motivation for sending
remittances. The initial work carried on remittances, by (Johnson and Whitelaw,
1974) explained such objective of the remittances. (Lucas and Stark, 1985)
explained that surely the most apparent motive for remitting was pure altruism in
which migrant considered the care for the family members at home. The utility
derived from such type of objective of the remittances was introduced. This utility
function explained that migrant’s utility was also the function of the consumptions
of his family members. In this way, utility function explained the highest degree of
the altruistic model of remittances. Indeed, this appeared the single idea in the
fundamental literature of the remittances.
The freshest theories of remittances were in the view that self- interest motives of
remittances nevertheless be the heart of the family as unit consideration. Such
type of literature about the remittances was at rest on the family as these
theories watch the family as a unit participating in business activities. The family
was able to enter into the optimization behaviors due to consideration of family
as unit. Several types of businesses and contracts were possible due to family
unification and a variety of self-centered models of remittances were explained
with the help of family as unit. (Lucas and Stark, 1985) further explained that
migrants might start investments projects while he was away from home due to
75
the consideration of family as complete unit because his family members could
look after his business without his stay in the country.
Stark (1991), and Gubert (2002) were opined about the functioning of the family
as an insurance company because these members were provided help and
protection in case of income decline from domestic resources with the help of
diversification of the sources for income generation. Poirine, (1997) suggested
that family worked like a bank to finance the expenditure of migration for its
members. Therefore, the migrant worker sent remittances back to family
members at home to repay the loans. They also remitted more loans for the
interests and welfare of other individual family members. This might be
considered a most consistent fact to get rid from the empirical literature written
on dimension and factors of the remittances. Many researchers exploring the
dimensions of remittances confirmed the altruistic behavior and very few
including (Agarwal and Horowitz, 2002) experienced altruism exchange with the
family planning. Lucas and Stark, (1985) found such evidence in the support of
self-interest behavior in case of Botswana and Agarwal and Horowitz (2002),
provided evidence in the support of altruism in Guyana.
76
3.5 Hypothesis Formulation
The review of different theories of remittances and exploring the economic
history of Pakistan were found helpful for selecting the relevant variables and the
construction of the econometric model to estimate the economic determinants of
international remittances and migration in Pakistan. The remittances interplayed
with economic, social and cultural factors that determined the scale and type of
remittances and their impact on the economy. An economic framework for
conceptualizing factors affecting the volume of the remittances in the economy
has been developed in the light of the theories of remittances. The theory of
migration has already been explained in the previous paragraph. The individual
migrant characteristics and migration problems were not addressed in this
research study. Given the scale of transfers, remittances could potentially be
determined by the economic and social conditions of the home as well as the
host countries.
The researchers have established various hypotheses for estimating the
determinants of the remittances in any country. First, it was possible that the size
and volume of remittances was driven by the household characteristic, including
the socio-demographic factors. Second, macroeconomic determinants were
more important in estimating the size of the remittances as compared to the
microeconomic determinants. Third, it was possible that remittances were mostly
determined by the home country conditions. Fourth, it was possible that the host
77
country economic conditions dominate the decision to send the remittances.
Last, but not least, it was possible that decisions of emigrant to send remittances
include both home and host country conditions.
In light of above stated hypotheses, it was not possible to test the every
hypothesis. Therefore, in this research study, the hypothesis was developed that
the size and volume of remittances was determined by the economic conditions
of the home country at macro level. Thus, the macroeconomic determinants of
international remittances in Pakistan were analyzed based upon the home
country conditions in the present study. Going through the theories of migration
and remittances, it was concluded that the migration was the integral part of the
remittances. In the literature survey, number of studies found repeatedly
migration as one of the determinant of remittances. Keeping in view, the
importance of the international migration, the economic determinants of
international migration were also estimated in the present study.
78
Chapter 4
Data and Methodology
4.1 Introduction
The devising right methodology and appropriate techniques were considered as
the center of any research study. Without making a right choice for methodology
and analysis, the impact study was futile exercise. Therefore, very important and
significant thing in conducting any analytical study was to adopt a systematic and
appropriate technique. The research methodology was considered very important
in the evaluation of any research problem because it completely explained the
entire procedure adopted in the relevant research project. The research proposal
highlights about the collection, methods, description and analysis of the data
used in the research study. It aids the researcher in the allocation of his limited
resources by selecting the specific choices from alternative methods (Sekaran,
1992).
Thus, the major objective of this chapter was to explain various tools and
techniques employed in the research study along with selection of sample,
collection, analysis and interpretation of data relating to research problem under
investigation. Therefore, the methodology used for finding the macroeconomic
determinants of international remittances and migration in Pakistan was
explained in this chapter. The description of the data, source of data, and
79
methodology adopted for the analysis of the data were also discussed in this
chapter.
4.2 Source of Data
The time series data were used in estimating the macroeconomic determinants
of international remittances and migration in Pakistan for the period of 1973 to
2005 in this study. The included variables in this research study were; the volume
of the remittances (US $ million), GDP at constant factor cost at 1980-81 prices
(Rs. Million), Real Growth Rate of the economy, Unemployment rate in the
country, Exchange Rate in Rs/$, Daily Wage Rate of unskilled workers at Lahore,
Literacy Rate, Inflation rate, Number of migrants went to abroad from Pakistan
per annum and Spread Rate of banks in Pakistan. These data series were
obtained from the Federal Bureau of Statistics, Islamabad. The Federal Bureau
of Statistics, Islamabad published the fifty years statistics of Pakistan and
provided the time series data about all the important variables of Pakistan
Economy. The data about the exchange rate was taken from the State Bank of
Pakistan, Karachi. The statistics about the migration was taken from the Bureau
of Emigration and Overseas Employment, Ministry of Labour, Manpower and
Overseas Pakistan.
However, real remittances (Rs. Million) and real wage rate were calculated with
the help of exchange rate and GDP deflator. The Spread Rate of banks was
80
determined by the difference of the interest rate on advances and saving
accounts. These variables were transformed into log form at the stage of model
specification in order to smooth out the data. The description of variables used in
this research study was given as under:
LRREM= Log of real remittances (Rs. Million),
LRGDP= log of real GDP (Rs. Million at constant factor cost 1980-81),
LGR= Log of real growth rate,
LUR= Log of unemployment rate,
LRWR= Log of real wage rate,
LLR= Log of literacy rate,
LSP= Log of spread rate,
LINF = Log of inflation rate and
LNW = Log of number of workers migrated abroad
4.3 Methods and Procedures
The time series data was used to estimate relationships among the variables
included in the model, to forecasts the future value of these variables and to test
different hypotheses very frequently. It was assumed that the time series data
were stationary in the traditional time series econometric. The developments in
the time series econometrics over the time revealed that most of the time series
were not stationary and if time series were not stationary, then the use of the
81
OLS method to analyze such data was not appropriate (Thomas, 1997). Most of
the time series were trended over time and relation among these series might
produce significant results with high R2, these results were termed as spurious or
meaningless (Granger and Newbold, 1974). It was the fact that many
econometricians used standard statistical methods developed for stationary data
series without checking the stationary of the data. Box and Jenkins (1970, 1976),
formulated regressions in which the time series were expressed in first
differences to resolve the problem of spurious results. It was assumed in this
approach that non-stationary data could be converted into stationary series by
taking differences repeatedly. However, this process of differencing resulted in
loss of valuable information among the series. Therefore, it was the main
disadvantage of this approach (Davidson et al., 1978).
The latest techniques of co-integration and error correction mechanism were
used in this research study to avoid the spurious results and overcome the
problem of the loss of valuable information of the used data. Granger (1981) and
Engle and Granger (1987) have introduced the concept of co-integration.
Econometric techniques based upon the co-integration considered the long run
relationships among the different variables included in the model. The co-
integration explained that an individual data series might move without specified
pattern but when this series was paired with another series, the pairs tended to
move together over time and showed relationship among them. The integrated
time series were modeled using a dynamic error correction mechanism under
82
certain conditions. This model was an extension of the traditional partial
adjustment model and has taken into account the dynamics of short-run
adjustment towards long-run equilibrium. The error correction model was based
upon the Granger theorem, which stated that if time series were found co-
integrated, then these series exhibited long run relationship among them (Engle
and Granger, 1987).
4.4 Stationary, Non-stationary and Order of Integration
It was assumed that the time series used in the study was stationary. A series Yt
was said to be stationary if the following conditions were fulfilled for all values of t
and t = I, 2, …, T
E (Yt) = µ (constant mean) (4.1)
Var (Yt) = E (Yt – µ)2 = σ2 (constant variance) (4.2)
cov (Yt, Yt+s) = cov (Yt, Yt-s) = γs (covariance depends on s, not t) (4.3)
Equations (4.1) and (4.2) stated that the series has a constant mean and
variance, while equation (4.3) stated that the covariance between any two values
of Y from the series (i.e., auto covariance) depended only on the difference part
in time between those two values (s) and not on the point in time (t). In equation
(4.3), ‘s’ indicated the difference part of time between two consecutive values of
the time series Y and t denoted the time period. Thus, a series was stationary if
83
its mean, variance and covariance all were found independent of time. A series
was non-stationary if it’s mean, variance or covariance were the function of time.
A stationary series has a tendency to move toward its mean and to fluctuate
around mean with a constant range, while a non-stationary series has a changing
mean at different points in time and its variance changes with the sample size
(Griffiths et al, 2001).
The number of times a series needed to be differenced for achieving stationary
was concerned with the number of unit roots contained in the series. The order of
integration equals to the number of unit roots a series contained, or a number of
differencing operations it took to make the series stationary. For example, a
series contained d unit roots and was integrated of order d, denoted by I (d), If it
became stationary after differencing d times (Dickey and Pantula, 1987).
4.5 The Concept of Co-integration
The co-integration explained the basic ideal that the included variables in the
model exhibit a long run relationship among the variables. If there was a long run
relationship among the variables, then divergence from the long run equilibrium
path was bounded and the variables were said to be co-integrated. Two
conditions were said to be satisfied for co-integration. First, the series included in
the model were co-integrated of the same order and second, a linear
combination of the variables exists which was integrated to an order lower than
84
individual variables. For example, if the variables become stationary after
differencing once, i.e., are I (1), then the error term from the co-integrating
regressions was stationary, i.e., I (0) (Hansen and Juselius, 1995). Suppose that
co-integrating regression was:
Yt = α + βXt + ut (4.4)
If the series Yt and Xt were both I (1) and the error term ut was I (0), then the
series were co-integrated of order I (1, 0). The β measured the equilibrium
relationship between the series Y and X and ut was the departure from the long
run equilibrium path in equation (4.4). The economic interpretation of co-
integration was that if in the long run two series Yt and X t or more series were
expressing an equilibrium relationship, then even though Yt and X t themselves
were non-stationary, they would move together closely over time and the
difference between these was constant (i.e., stationary). Therefore, the term co-
integration represented a long-run equilibrium to which a system of series moved
with time and ut may thus be interpreted as the disequilibrium error i.e., the
extent to which relationship deviated from equilibrium (Johansen, 1994).
4.6 Order of integration
Numbers of test were used to find the order of co-integration of series in
econometrics literature but most commonly was unit roots test. It was formulated
85
and developed by the augmented Dickey – Fuller (Dickey and Fuller, 1981).The
following equation was estimated in this test.
k ∆Yt = β1 + β2T + β3Yt-1 + Σ θi ∆ Yt-i + ut (4.5) i = 1
Where: Yt was the series for which, the order of integration was found, T was the
time trend and ut was white noise residuals. The number of lags, k, was unknown
and Lagrange Multiplier test (LM test) from general-to-specific with maximum k =
13 at 95 percent confidence level was used to know these lags (Holden and
Perman, 1994). It was assumed in LM test that there was one unit root in the
process (Dickey, Bell and Miller, 1986). However, if there was more than one unit
roots, the standard testing procedure was to test first for a unit root in the levels
of series Yt; then, if the null hypothesis was not rejected then to test the first
difference i.e. ∆Yt for the presence of a second unit root and so on. This testing
procedure from lower to higher orders of integration continued until the null
hypothesis was rejected. However, Dickey and Pantula (1987) argued that it was
not a valid procedure since the ADF – tests were based on the assumption of
only one unit root. Thus, it was inappropriate to test for unit roots from lower to
higher orders when there were more than one unit root in a series and sequential
method of Dickey and Pantula (1987) were followed. The null hypothesis of three
unit roots was assumed and tested. Null hypothesis of two unit roots was tested
on the rejection of three-unit roots hypothesis and so on until the null was
accepted. The ADF – test for three unit roots was formulated as:
k
86
∆3Yt = α + ץYt-1 Σθi∆3 Yt-i+ ut (4.6) i=1
The stationary properties and the test of unit root in this research study were
performed by the augmented Dickey – Fuller (ADF) test. The reported statistics
in this study were confined to the DF (ADF) approach in levels, because in most
instances the DF (ADF) led to the same conclusions. The null hypothesis of a
unit root could not be rejected at the 5 percent level in the series because of
integrated of order one i.e. I(1).
4.7 Test of Co-integration
Two formal approaches were commonly employed to observe the presence of
co-integration among included series in the model. These approaches were: the
augmented Dickey-Fuller residual-based test approach proposed by Engle and
Granger (1987) and the Johansen’s Full Information Maximum Likelihood (FIML)
approach (Johansen and Juselius, 1990). The co-integration of the included
variables in the model and the residual term was tested in the residual-based
tests. It was the drawback of the residual-based approach that it included only
two variables in the model. In this way, there will be only one co-integrating
vector. However, if the number of variables were increased in the model, there
was more than one co-integrating vector. In this case, the use of residual-based
approach was not appropriate. Consequently, Johansen’s method, which allowed
the estimation of all the possible co-integrating relationships existing among the
87
variables, was used. Its advantage was that co-integrating vectors could be
determined empirically. However, this method could not be compared with the
residual-based single equation test and system-based tests directly because they
were grounded within different econometric methodologies.
4.8 Co-integrating Vectors
Two likelihood ratio (LR) tests were used to find out the presence of a single co-
integrating vector. These tests were trace test statistic and maximal-eigen value
test. The null hypothesis of at most r co-integrating vectors against the alternative
that it was greater than r could be tested under the trace test statistic and was
explained as:
p λ trace (r) = - T Σ ln (1- λ i) (4.10)
I = r+1
The r co-integrating vectors could be tested in the null hypothesis as compared
to the alternative that was r+1 according to the maximal-eigen value test and was
explained as:
λ max (r, r + 1) = - T ln (1- λ t+1) (4.11)
The critical values of these tests have been derived by Monte Carlo Simulations
and tabulated by Johansen (1988) and Osterwald- Lenum (1992). Harris (1995,
88
p.89) noted that between these two LR tests for co integration, the trace test
showed more robustness.
In this research study to estimate the determinants of remittances and migration,
the adjusted likelihood ratio (LR) statistics (Sims, 1980) was used to find the
value of k i.e order of the VAR. The conclusion of the adjusted LR were drawn on
the AIC and SBC criteria. The issue of the estimated coefficients being the long-
run elasticities in the co-integrating vectors was not clear. This was true when
there were only two variables in the co-integrating vector, but when there were
more variables in the model, the dynamics of the VAR raised some doubts about
their interpretation (Lutkepohl, 1993). Nevertheless, some authors Hallam and
Zanoli, (1993) interpreted the estimated coefficients as the long-run elasticities.
In this study the estimated coefficients were interpreted as long-run elasticities of
the remittances and migration because the model was specified in the
logarithmic form.
4.9 Error Correction Mechanism
The error correction mechanism stated that a proportion of disequilibrium in a
system (model) from one period was corrected in the next period in the said
system (Engle and Granger, 1987). The error correction mechanism was first
used and explained by (Sargan, 1964). It was mentioned that traditional
econometric estimation assumed that the time series were stationary i.e., I (0).
89
However, most time series were not stationary and were integrated of a certain
order. Mainly series were either trend-stationary (TS) or differenced stationary
(DS). A trend stationary series was rendered stationary by including a trend term
and a differenced stationary series was rendered stationary by differencing. But
the process of differencing or transforming a time series to stationary series also
leads to the loss of some important long-run information which otherwise was
required to be retained for estimation. The error correction models, however,
solved the problem of integrated data series using traditional estimation methods
without losing long-run information. The basis of error correction model (ECM)
was the Granger representation theorem. This theorem represented that if the
variables were co-integrated, there was a long-run relationship between them
and was described by the error correction model. An ECM involving the variables
Y and X in its simplest form was expressed as:
∆Y t = α ∆X t - ø (Yt-1 - δXt-1) + µ t (4.12)
Where: µ t was the disturbance term with zero mean, constant variance, and zero
covariance. Parameter α taken into account the short-run effect on Y of the
changes in X, while δ measured the long-run equilibrium relationship between Y
and X i.e.
Yt = δ Xt + µt (4.13)
90
Where: Yt-1 - δYt-1 = µt-1 measured the divergence from long-run equilibrium.
Finally, ø measured the extent of error correction by adjustment in Y and its
negative sign indicated that the adjustment was in the direction, which restored
the long-run relationship (Hallam and Zanoli, 1993). Engle and Granger (1987)
proposed a two-stage procedure for estimating (4.12). In the first stage, the static
long-run co-integration regression (4.13) was estimated to test co-integration
between the two variables. If co-integration existed, then the lagged residuals
from the static co-integrating regression (4.13) was used as the error correction
term in the error correction model (4.12) to estimate the short run equilibrium
relationship between the variables in the second stage. The validity of ECMs
depended upon the existence of a long run or equilibrium relationship among the
variables.
The ECM has several advantages. It contained a well-behaved error term and
avoided the problem of autocorrelation. It allowed a consistent estimation of
parameters by incorporating both short-run and long run effects. The major
advantage of ECM was that all variables were stationary in the model. It ensured
that no information on the levels of variables was ignored by the inclusion of the
disequilibrium terms because ECM was formulated in terms of first difference. It
eliminated trends from the variables and resolved the problem of spurious results
(Granger and Newbold, 1978).
91
The use of vector correction models (VECM) could solve the endogeneity
problem among remittances and its macroeconomic determinants. The causality
ran both the ways. For example, the remittances caused GDP and GDP caused
remittances (CV, Silva, 2005). This endogeneity appeared to be present because
while some studies have found that macroeconomic variables in the home
country affected the remittances, other studies have found that remittances
affected the macroeconomic variables in the home country.
4.10 Modeling Strategy
The following important steps were involved in explaining the model strategy
used in this research study.
(i) Augmented Dickey_ Fuller (ADF) unit root test was employed to find
out the order of integration of the series used in the research study.
The stationary of these series was determined with the presence of a
unit root.
(ii) The log form of the variable was suggested (Schrooten, 2005) for
model specification for smooth pasting of data and on the basis of the
more accurate estimates of the required parameters.
(iii) Variables used in the research study were found integrated of the
same order i.e. I (1), therefore Johansen and Juselius (1990)
maximum likelihood estimation approach was used to test the co-
92
integration.
(iv) Moreover, the Engle Grange techniques based on the residual test
were also applied to test the co-integration.
(v) The lag k of the VAR model was specified before undertaking the co-
integration test. The VAR model treated every variable within the
system and the equations of these variables were estimated with
lagged values.
The variables used in the research were found co-integrated, therefore, their long
run relationship was estimated via ordinary least square (OLS) method. The
Vector Error Correction Model (VECM) was used for the estimation of short run
adjustment.
The two econometric models were specified in the research study with the guide
line provided by the outlined model strategy.
(i) Real Remittances Model
Real Remittances Model was used to obtain the economic determinants of
international remittances in Pakistan. The model parameters were defined
keeping in view the empirical literature. The functional form of the real
remittances model was:
RREM = F (RGDP, GR, UR, LR, SP, RWR).
93
The estimating econometric model was:
LRREM = β0 + β1LUR + β2LRWR + β3LSP + β4LGR + β5LLR + β6LRGD + et
(ii) Migration Model
Migration Model was used to identify the economic determinants of international
migration in Pakistan. The parameters of the model were specified based on
previous studies and theoretical literature. The functional form of the migration
model was:
NW = F (INF, RREM, UR, RWR);
The econometric model for the estimation was:
LNW = β0 + β1LUR + β2LRWR + β3LINF + β4LRREM + et
94
Chapter 5
Unit Root Test
5.1 Introduction
The time series moved upwards or downwards without any specified patterns
were known as random walks and the series showing a definite trend either
upwards or downwards were known as random walks with a drift. Considered AR
(1) process
Yt = α + ρYt-1 + εt (5.1)
The AR (1) process in (5.1) was stationary if ⎜ρ ⎜< 1 and the AR (1) process in
(5.1) was non-stationary if ⎜ρ ⎜= 1. Further, the AR (1) process in (5.1) reduced to
a non-stationary random walk series if α = 0 and ⎜ρ ⎜= 1. The produced series
was non-stationary and called a random walk with drift in case of α ≠ 0 and ⎜ρ ⎜=
1 (Griffiths et al., 2001).
The time series data was used in this research study to find out the determinants
of the real remittances and migration in Pakistan. Therefore, it was necessary to
know about the properties of the time series data used in the study. The
stationary of time series data was explored with the help of informal and formal
95
approaches available in the econometrics literature including the Auto Correlation
Functions and Unit Root Tests respectively. However, the results of formal
approaches like the Unit Root Test were reported in the study. The graphs of
these series were also drawn to ascertain the pattern of these series over the
time.
5.2 Unit Root Test
The stationarity of series used in the study was determined with the estimation of
a unit root. Dickey Fuller (DF) unit root test might be estimated from the following
forms of equations.
1. Without constant and trend; Yt = ρYt-1 + ε t (5.2)
2. With constant; Yt = α + ρYt-1 + ε t (5.3)
3. With constant and trend; Yt = α + ρYt-1 + βT+ ε t (5.4)
Where: Yt was the relevant time series, α was constant (Intercept), T was time
trend and ε t was the residual term. This test was carried out on levels and first
differences. The null hypothesis was that the variable has a unit root i.e. ρ = 1
against the alternative hypothesis having no unit root. If null hypothesis was
accepted, then Yt series was non- stationary time series. The null and alternative
96
hypotheses were stated as: H0: ρ = 1 and H1: ρ ≠ 1 and the decision rule for
testing these hypotheses was framed as: The null hypothesis was rejected and
concluded that unit root not existed if ADF* > t critical value and null hypothesis
was not rejected and concluded that unit root existed in case of ADF < t critical
value.
Equations 5.2, 5.3 and 5.4 might be transformed into differenced equations by
subtracting Yt-1 on both sides of these equations and were given in the following
forms to obtained results.
4. Without constant and trend; ∆Yt = δYt-1 + ε t (5.5)
5. With constant; ∆Yt = α + δYt-1 + ε t (5.6)
6. With constant and trend; ∆Yt = α + δYt-1 + βT+ ε t (5.7)
Where: ∆Yt = Yt - Yt-1 and δ = ρ - 1. The null and alternative hypotheses
formulated in the light of new regression form equations were H0: δ = 0 (Unit
Root) and H1: δ≠ 0 (No Unit Root). Augmented Dickey Fuller (ADF) unit root test
was based on the following regression equation.
∆Yt = α + βT + δYt-1 + γ i ∆Yt-i + ε t (5.8)
97
Various regression forms were estimated to find the presence of the unit root in
the data series and results of the unit root test were reported in Table 5.1 through
Table 5.4. It was found that all the time series used for estimating the
determinants of international remittances and migration in Pakistan were non-
stationary and their order of integration was one i.e. I (1). The variables in the
study were transformed into log form while specifying the model for carrying out
the research.
Table 5.1: Unit Root tests using Augmented Dickey-Fuller Method Variables Without trend With trend
RREM -2.07 -2.14
RGDP 2.60 1.23
GR -2.91 -2.71
UR -.74 -2.32
RWR -2.91 -2.85
LR 1.03 -1.40
SP -1.22 -2.53
INF -2.69 -2.18
NW 2.3 -0.59
• Critical value for the augmented Dickey-Fuller statistic with intercept and without trend was –2.96 (p = 0.05 per cent) • Critical value for the augmented Dickey-Fuller statistic with intercept and trend was –3.56 (p = 0.05 per cent)
The reported statistics in this study was, however, confined to the ADF (AD)
approach including intercept with and without trend. AD and ADF up to level
98
three were calculated to know the real situation. Informal test like auto correlation
function and partial correlation function were also performed but not reported.
The results of the unit root tests of the series RREM, RGDP, GR, UR, RWR, LR,
INF, NW and SP were reported in Table 5.1. These series were found non-
stationary on the basis of the augmented Dickey-Fuller statistic. The null
hypothesis was not rejected because t calculated was greater than the ADF
critical value. Thus, the unit root was existed and confirmed that the series were
non-stationary.
Table 5.2: Unit Root tests using Augmented Dickey-Fuller Method Variables Without trend With trend
LRREM -2.94 -2.56
LRGDP -.12 -1.43
LGR -2.02 -2.93
LUR -1.51 -2.69
LRWR -2.80 -2.73
LLR -.30 -1.86
LSP -2.04 -2.79
LINF -2.68 -2.73
LNW .008 -2.54
• Critical value for the augmented Dickey-Fuller statistic with intercept and without trend was –2.96 (p = 0.05 per cent) • Critical value for the augmented Dickey-Fuller statistic with intercept and trend was –3.56 (p = 0.05 per cent)
99
During the model specification, log form showed robust results. Therefore, log
transformation was applied to all the data series to ascertain the stationary of the
series. The results of the log form of the variables were reported in the Table 5.2.
It was concluded in the light of results that all series remained non-stationary
after log transformation.
Table 5.3: Unit Root tests using Augmented Dickey-Fuller Method Variables Without trend With trend
∆RREM -3.98 -3.90
∆RGDP .72 -.59
∆GR -5.16 -5.14
∆UR -4.16 -4.06
∆RWR -3.78 -3.64
∆LR -1.91 -2.01
∆SP -5.05 -4.99
∆INF -5.87 -6.26
∆NW -2.98 -2.86
• Critical value for the augmented Dickey-Fuller statistic with intercept and without trend was –2.96 (p = 0.05 per cent) • Critical value for the augmented Dickey-Fuller statistic with intercept and trend was –3.57 (p = 0.05 per cent) • ∆: Indicate the first difference of the series.
The unit root test was applied after differencing once each series and results
were reported in Table 5.3. ∆RGDP and ∆LR series were found non-stationary
on the basis of these results. All other series were stationary after differencing
100
once. In order to solve this problem, log transformation was applied after
differencing once and results were reported in Table 5.4. Therefore, log
transformation was justified on two grounds. It was more appropriate for model
specification and it also solved the problem of stationary. The results in Table 5.4
indicated that all the series were stationary. Therefore, it was concluded that all
series used in this research study were of the same order i.e. I (1).
Table 5.4: Unit Root tests using Augmented Dickey-Fuller Method Variables Without trend With trend
∆LRREM -3.58 -361
∆LRGDP -2.98 -3.61
∆LGR -5.30 -5.25
∆LUR -3.99 -5.25
∆LRWR -3.35 -3.28
∆LLR -3.31 -4.25
∆LSP -5.38 -5.32
∆LINF -4.62 -4.66
∆LNW -2.98 -3.89
• Critical value for the augmented Dickey-Fuller statistic with intercept and without trend was –2.96 (p = 0.05 per cent) • Critical value for the augmented Dickey-Fuller statistic with intercept and trend was –3.57 (p = 0.05 per cent) • ∆ : Indicates the first difference of the series.
101
5.3 Graphic presentation of the Series
In this section, the graphs of all the time series data were drawn to ascertain the
pattern of series over time. The graphs of time series were helpful for assessing
the properties of the series, as the conditions for stationary were difficult to grasp.
The graph of real remittances (RREM) in Pakistan was shown in figure 5.1. The
time period from 1973 to 2005 in years was taken on horizontal axis and real
remittances (in Rupees) were shown on vertical axis respectively. The graph of
real remittances represented as random walk with drift because this graph
showed a definite upward trend. Therefore, real remittances were non stationary
series.
Real Remittances
01000020000300004000050000
1970 1980 1990 2000 2010
Years
Real
Rem
ittan
ces
Series1
Figure 5.1: Real Remittances in Pakistan
102
The graph of real gross domestic product (RGDP) in Pakistan was shown in
figure 5.2. The time period from 1973 to 2005 in years was taken on horizontal
axis and real domestic product (in Rupees) of the country was shown on vertical
axis respectively. The graph of real GDP looked like random walk with drift
because this graph showed a definite upward trend. Therefore, real GDP were
non stationary series.
Real GDP
0200,000400,000600,000800,000
1,000,000
1970 1980 1990 2000 2010
Years
Real
GDP
Series1
Figure 5.2: Real GDP in Pakistan
103
The graph of growth rate (GR) in Pakistan was shown in figure 5.3. The time
period from 1973 to 2005 in years was taken on horizontal axis and growth rate
of the economy was shown on vertical axis respectively. The graph of real growth
rate looked like random walk. Therefore, real growth rate were non stationary
series.
Real Growth Rate
02468
10
1970 1980 1990 2000 2010
Years
Gro
wth
Rat
es
Series1
Figure 5.3: Real Growth Rate in Pakistan
104
The graph of real wage rate (RWR) in Pakistan was shown in figure 5.4. The time
period from 1973 to 2005 in years was taken on horizontal axis and real wage
rate (in Rupees) were shown on vertical axis respectively. The graph of real
wage rate looked like random walk. Therefore, real wage rate were non
stationary series.
Real Wage Rate
0.0010.0020.0030.0040.0050.00
1970 1980 1990 2000 2010
Years
Rea
l Wag
e R
ate
Series1
Figure 5.4: Real Wage Rate in Pakistan
105
The graph of unemployment rate (UR) in Pakistan was shown in figure 5.5. The
time period from 1973 to 2005 in years was taken on horizontal axis and
unemployment rate in Pakistan was shown on vertical axis respectively. The
graph of unemployment rate looked like random walk with drift because this
graph showed a definite upward trend. Therefore, unemployment rate were non
stationary series.
Unemployment Rate
02468
10
1970 1980 1990 2000 2010
Years
Unem
ploy
men
t R
ate
Series1
Figure 5.5: Unemployment Rate in Pakistan
106
The graph of literacy rate (LR) in Pakistan was shown in figure 5.6. The time
period from 1973 to 2005 in years was taken on horizontal axis and literacy rate
in the country was shown on vertical axis respectively. The graph of literacy rate
looked like random walk with drift because this graph showed a definite upward
trend. Therefore, literacy rate were non stationary series.
Literacy Rate
0
20
40
60
1970 1980 1990 2000 2010
Years
Lite
racy
Rat
e
Series1
Figure 5.6: Literacy Rate in Pakistan
107
The graph of spread rate (SP) in Pakistani banks was shown in figure 5.7. The
time period from 1973 to 2005 in years was taken on horizontal axis and spread
rate of banks in Pakistan was shown on vertical axis respectively. The graph of
the spread rate looked like random walk. Therefore, spread rate were non
stationary series.
Bank Spread Rate
02468
10
1970 1980 1990 2000 2010
Years
Bank
Spr
ead
Rate
Series1
Figure 5.7: Bank Spread Rate in Pakistan
108
The graph of number of workers (NW) migrated from Pakistan to other countries
per annum was shown in figure 5.8. The time period from 1973 to 2005 in years
was taken on horizontal axis and number of workers migrated from Pakistan per
annum were shown on vertical axis respectively. The graph of the number of
workers looked like random walk with drift because this graph showed a definite
upward trend. Therefore, flows of migration rate were non stationary series.
out flow of migrant workers per year
0
50000
100000
150000
200000
250000
1970 1980 1990 2000 2010
Years
Num
ber o
f wor
kers
Series1
Figure 5.8: Outflow of Migrant Workers from Pakistan per annum
109
The graph of inflation (INF) in Pakistan was shown in figure 5.9. The time period
from 1973 to 2005 in years was taken on horizontal axis and inflation rate in
country was shown on vertical axis respectively. The graph of the inflation rate in
Pakistan looked like random walk. Therefore, inflation rate were non stationary
series.
Inflation Rate
010203040
1970 1980 1990 2000 2010
Years
Infla
tion
Rate
Series1
Figure 5.9: Inflation Rate in Pakistan
110
Chapter 6
Co-integration Analysis
6.1 Introduction
The real remittances model and migration model were specified in this research
study to investigate the economic determinants of international remittances and
migration in Pakistan. The co-integration among the variables used in these
models was found out using two approaches. These approaches were
Augmented Dickey-Fuller (ADF) residual-based approach designed by Engle and
Granger (1987) and the Johansen’s Full Information Maximum Likelihood (FIML)
approach planned by Johansen and Juselius (1990). First of all the co-integration
of the real remittances model was presented with brief introduction of the co-
integration. Later on, the co-integration among the variables used in migration
model was determined.
6.2 Co-integration
The basic idea underlying the concept of co-integration was to identify the
equilibrium or a long run relationship among the variables. For instance, if series
of the variables Y t and X t were stationary after differencing once i.e. I (1) and
error term u t from the co-integrating regressions was stationary, i.e. I (0), then
the series Y t and X t were co-integrated of order I (1, 0), (Johansen and Juselius,
111
1995). As already pointed out the co-integrating equation was
Y t = α + βX t + u t (6.1)
If the series Yt and Xt were both I (1) and the error term ut was I (0), then the
series Yt and Xt were co-integrated of order I (1, 0). In equation (6.1), β measures
the equilibrium relationship between the series Yt and Xt and ut was the
departure from the long-run equilibrium path. In other words, the series of at least
two variables were co-integrated if a linear combination of these variables existed
which was integrated to an order lower than individual variables.
6.3 Co-integration Analysis of Remittances Model
In the first step, the co-integration among the variables used in the real
remittances model was analyzed based upon the residual test. All the variables
in the real remittances model were non-stationary and were of integrated order
one i.e. I (1) in equation (6.2).
LRREM = β0 + β1LUR + β2LRWR + β3LSP + β4LGR + β5LLR + β6LRGD+e t (6.2)
The equation (6.2) was estimated by OLS method to apply the residual test.
ADF-test was conducted upon the residual et obtained from equation (6.2). The
calculated ADF value of et was -5.99 at .05 per cent level without trend and was -
112
5.89 at .05 per cent level with trend where as the corresponding critical values of
these ADF were -2.96 and -3.57 respectively. It showed that calculated values
were less than the critical values. Therefore, it was concluded that residual et
was stationary having integrated order zero i.e. I (0) and other variables including
depend as well as explanatory in the model were integrated order one i.e. I (1).
Therefore, co-integration among the variables in equation (6.2) was existed. The
co-integration among the variables used in the remittances model was also
determined by employing the Johansen’s Full Information Maximum Likelihood
(FIML) approach to overcome the problems encountered in the residual test. For
this purpose, the order of VAR was found out and described in the following
section.
6.3.1 Order of VAR
It was important to find the order of Vector Auto regression (VAR) model before
estimating the Johansen’s co-integration tests. Therefore, the specification of the
significant order of lags (p) of the Vector Auto regression (VAR) model was
estimated. The order of lags considered to be p = 1 for the annual data (Pesaran
and Pesaran, 1997).
113
Table 6.1: Selecting the Order of VAR for the Real Remittances Model List of variables included in the unrestricted VAR:
LRREM LUR LRWR LSP LGR LLR LRGDP
List of deterministic and/or exogenous variables
CONSTANT
Order AIC SBC Adjusted LR test
10 -6.51 -16.16 --------
9 -5.60 -14.68 .04[.83]
8 -5.04 -13.56 .28[.87]
7 -4.09 -12.05 .31[.96]
6 -5.36 -12.74 1.49[.83]
5 -5.59 -12.40 2.13[.83]
4 -5.18 -11.42 2.43[.88]
3 -4.25 -9.93 2.47[.93]
2 -3.47 -8.58 2.59[.96]
1 -3.48 -8.02 3.11[.96]
0 -9.99 -13.97 7.03[.72]
Note: p – values in the parentheses.
AIC=Akaike Information Criterion SBC=Schwarz Bayesian Criterion
Equation 6.3 was used to specify the order of lags of unrestricted vector auto
regression (VAR) for VECM modeling in Johansen procedure.
Z t = A t Z t -1 + … + A k Z t -K + µ t (6.3)
114
Where: Zt was (n×1) and each of the Ai was a (n×n) matrix of parameters.
Adjusted LR tests with a maximum of ten lags were carried out to determine the
order of VAR. The results of the real remittances model were given in Table 6.1.
Adjusted LR test statistics, Schwarz Bayesian Criterion (SBC) and Akaike
Information Criterion (AIC) were provided in the Table 6.1 for the selection of the
order of the VAR. The order of VAR was selected as p = 1 on the basis of SBC to
avoid over-parameterization of a time series.
6.3.2 Vectors of Co-integration
The vector of co-integration represented a stationary linear combination of non-
stationary variables. There might be more than one co-integrating vectors of the
relevant variables used in the research. In other words, the series of at least two
variables of the same order has at least one co-integrating vector if a linear
combination of these variables existed, which was integrated to an order lower
than individual variables (Johansen and Juselius, 1990). Therefore, it was
necessary to test the presence of co-integration and to determine the number of
co-integrating vectors among the series used in the real remittances model. The
unrestricted intercept and no trend model were used to find the co-integrating
vectors in the Johansen co-integration model. The results of the co-integration
were provided in the Table 6.2. Six co-integrating vectors were selected on the
foundation of the Eigen Value Test. Therefore, it was concluded that the included
variables in the real remittances model were co-integrated.
115
Table 6.2: Johansen Co-integration Results for Real Remittances Model Relationship Hypotheses Eigen values
Critical values
H0: r Ha: r
LRREM, LSP
LRWR, LUR,
LG, LRGDP,
LLR
0
1
2
3
4
5
6
1
2
3
4
5
6
7
157.73
64.86
44.18
31.33
23.51
21.25
8.11
46.47
40.53
34.40
28.27
22.04
15.87
9.16
The critical values were given (p = 0.05 per cent) levels for Co-integration.
Table 6.3 provided the co-integration results estimated on the base of Trace
Value Test with unrestricted intercepts and no trends. Six co-integrating vectors
were also selected on the basis of Trace Value Test and it was concluded that
the included variables in the real remittances model were co-integrated. The
results of co-integration presented in Tables 6.2 and Table 6.3 revealed that real
remittances model was having six co-integrating vectors on the basis of Eigen
Value Test as well as on the basis of Trace Value Test.
116
Table 6.3: Johansen Co-integration Results for Real Remittances Model Relationship Hypotheses Trace Values
Critical values
H0: r Ha: r
LRREM, LSP
LRWR, LUR,
LG, LRGDP,
LLR
0
1
2
3
4
5
6
1
2
3
4
5
6
7
350.98
193.25
128.39
84.21
52.87
29.36
8.11
132.45
102.56
75.98
53.48
34.87
20.18
9.16
The critical values were given (p = 0.05 per cent) levels for co-integration.
6.4 Co-integration Analysis of Migration Model
The co-integration among the variables used in the migration model was
analyzed using the both approaches as in case of real remittances model. In the
first step co-integration among the variables used in the migration model was
analyzed based upon the residual test. The variables included in the migration
model were non-stationary and were of integrated order one i.e. I (1) in equation
(6.4).
LNW = β0 + β1LUR + β2LRWR + β3LINF + β4LRREM + et (6.4)
117
The equation (6.4) was estimated by OLS method to apply the residual test.
ADF-test was conducted upon the residual et obtained from equation (6.4). The
calculated ADF value of et was -4.86 at .05 per cent level without trend and was -
4.52 at .05 per cent level with trend where as the corresponding critical values of
these ADF were -2.96 and -3.57 respectively. It showed that calculated values
were less than the critical values. Therefore, it was concluded that residual et
was stationary having integrated order zero i.e. I (0) and other variables in the
model were integrated order one i.e. I (1). Therefore, co-integration among the
variables in equation (6.4) existed. The co-integration among the variables used
in the migration model was also determined by employing the Johansen’s Full
Information Maximum Likelihood (FIML) approach. For this purpose, the order of
VAR was found out and described in the following section.
6.4.1 Order of VAR
The estimation of the order of Vector Auto regression (VAR) model was
important before the application of the Johansen’s co-integration tests.
Therefore, the order of lags (denoted by p) of the Vector Auto regression (VAR)
model were specified and estimated here in this section. Equation 6.5 was used
to specify the order of lags of unrestricted vector auto regression (VAR) for
VECM modeling in Johansen procedure.
Z t = A t Z t -1 + … + A k Z t -K + µ t (6.5)
118
Where: Z t was (n×1) and each of the A i was a (n×n) matrix of parameters.
Adjusted LR tests with a maximum of ten lags were carried out to determine the
order of VAR.
Table 6.4: Selecting the Order of VAR for the Migration Model List of variables included in the unrestricted VAR:
LNW LRREM LUR LINF LRWR
List of deterministic and/or exogenous variables
CONSTANT
Order AIC SBC Adjusted LR test
10 7.49 -1.02 --------
9 8.49 .54 .005[.95]
8 5.32 -2.06 2.89[.24]
7 4.87 -1.93 3.91[.27]
6 .83 -5.40 7.41[.12]
5 1.81 -3.86 7.42[.19]
4 2.15 -2.96 7.89[.25]
3 1.40 -3.14 9.10[.24]
2 1.96 -2.02 9.42[.31]
1 -1.38 -4.78 12.43[.19]
0 -1.81 -4.65 13.42[.20]
Note: p – values in the parentheses.
119
The results of the migration model were given in Table 6.4. Test Statistics and
Choice Criteria for Selecting the order of the VAR Model was based on 30
observations from 1976 to 2005 including p = 10 i.e. the order of VAR. Adjusted
LR test statistics, Schwarz Bayesian Criterion (SBC) and Akaike Information
Criterion (AIC) were also provided in the Table 6.4. The order of VAR was
selected as p = 1 on the basis of SBC to avoid over-parameterization of a time
series.
6.4.2 Vectors of Co-integration
The unrestricted intercept and no trend model were used to find the co-
integrating vectors in the Johansen co-integration model.
Table 6.5: Johansen Co-integration Results for Migration Model Relationship Hypotheses Eigen values
Critical values
H0: r Ha: r
LNW, LRREM,
LINF, LRWR,
LUR
0
1
2
3
4
1
2
3
4
5
39.58
32.23
19.57
8.61
3.44
34.40
28.27
22.04
15.87
9.16
The critical values were given (p = 0.05 per cent) levels for Co-integration.
120
The results of the co-integration based on Maximal Eigen Value of the Stochastic
were given in Table 6.5. Two co-integrating vectors were selected on the basis of
the Eigen Value Test. Therefore, it was concluded that the included variables in
the migration model were co-integrated.
The results of the co-integration based on Maximal Trace value of the Stochastic
Matrix were given in Table 6.6. Two co-integrating vectors were also selected on
the basis of Trace Value Test and it was concluded that the included variables in
the migration model were co-integrated.
Table 6.6: Johansen Co-integration Results for Migration Model Relationship Hypotheses Trace Values
Critical values
H0: r Ha: r
LNW, LRREM,
LINF, LRWR,
LUR
0
1
2
3
4
1
2
3
4
5
103.43
63.85
31.62
12.05
3.44
75.98
53.48
34.87
20.18
9.16
The critical values were given (p = 0.05 per cent) levels for co-integration.
The results of co-integration presented in Tables 6.5 and Table 6.6 revealed that
migration model was having two co-integrating vectors on the basis of Eigen
Value Test as well as on the basis of Trace Value Test.
121
Chapter 7
Economic Determinants of International Remittances
7.1 Introduction
The economic determinants of international remittances in Pakistan were
explored in this chapter. The literature review demonstrated that the remittances
were dependent upon unemployment rates, wage rates, literacy rates in home as
well as host countries. Similarly, remittances were also related with the economic
conditions of the sending and receiving countries. Therefore, the estimates of the
remittances were considered very important for the policy maker to expand the
remittances and formulate suitable policy. The micro and macro determinants
have an important bearing on the flow and quantum of remittances. In this study,
only macro determinants were explored and their long run and short run effects
were analyzed.
7.2 Long Run Relationship
The variables in the real remittances model were co-integrated exhibiting a long
run relationship. Their long run relationship was determined with ordinary least
square (OLS) method. The results of the OLS estimates were reported in Table
7.1. The first column of the table represents the variables (already defined) in log
form whereas C was the intercept in the first row. The estimated coefficients
were reported in 2nd column. The estimates were significant on the basis of t
122
statistics. No serial correlation among the residuals was detected on the basis of
D-W statistics.
The results in Table 7.1 explained that the real remittances were positively
related with the unemployment rate in the country. The coefficient of
unemployment rate was estimated 0.718 and was highly significant at 0.01
probability level. It was the partial elasticity of remittances with respect to
unemployment because remittances model was specified in the log form. It
meant that remittances were increased 0.718 per cent due to one per cent
increase in the unemployment rate of the country. It showed that when there was
unemployment in the country, people felt unease and frustrated in the country
and moved to foreign countries for their livelihood. In this way, migrant brought
more remittances in the country. However, the study did not go into the details of
the types of unemployment due to limitation of data. Schrooten, (2005) also
estimated the determinants of remittances and found unemployment rate as one
of the determinants of remittances for the former socialist countries. The results
of the present study were robust and consistent with the previous studies.
The real remittances were positively related with the real growth rate and real
GDP of the country. The estimated coefficients of these variables were 0.236 and
3.818 and were found significant at 0.10 and 0.01 probability levels respectively.
123
Table 7.1: Regression Results relating Real Remittances with Independent Variables in Pakistan Variable (Ln) Coefficient Std. Error t-Statistic Prob.
C -9.641 4.858 -1.984** 0.058
UR 0.718 0.346 2.072*** 0.049
RWR -2.164 0.498 -4.341*** 0.0002
SP -0.430 0.144 -2.984*** 0.006
GR 0.236 0.144 1.637* 0.114
LR -6.576 1.460 -4.504*** 0.0001
RGDP 3.818 0.794 4.807*** 0.0001
R2 0.83
Adjusted R2 0.79
Durbin-Watson 2.15
* Indicated that the coefficient was significantly different from zero at 0.10 probability level. ** Indicated that the coefficient was significantly different from zero at 0.05 probability level. *** Indicated that the coefficient was significantly different from zero at 0.01 probability level. Ln indicated that variables were in the logarithmic form.
The real growth rate and real GDP were the signs of prosperity in the country
and attracted more remittances in the country. The migrants tend to send more
remittances for investment motives, thus augmented the real GDP of the country.
Sakka and Mcnabb (1999) concluded that the amount of remittances was
affected positively by the gross domestic product (GDP) of the home country
during the estimation of the macroeconomic determinants of the remittances. The
results of the study also showed that the domestic GDP positively affected the
124
flow of remittances in both its current and lagged forms. However, Schrooten
(2005) showed that growth rate has little impact upon remittances in the short-run
and no effect in the long run. Several other researchers (Rahaman, 2003; Aydas,
2004; Schiopu and Seigfried, 2006) found a positive relationship between the
growth rate and the remittances. The results of this study were consistent with
other researchers. However, Chami et al. (2005) found a negative relationship
between the growth rate and the remittances.
The real remittances were found negatively related with the spread rate of banks
in Pakistan in this study. The estimated coefficient was –0.43 and was highly
significant at 0.01 probability level. This indicated that the migrant were not
motivated to keep their savings in Pakistani banks. The migrant deposited their
saved money in foreign banks due to large spread rate in Pakistan. However,
Schrooten (2005), found a positive relationship between the spread rate and the
remittances due to financial innovations in the socialist countries.
The real remittances were negatively related with the literacy rate in Pakistan.
The estimated coefficient of literacy rate was –6.576 and was found significant at
0.01 probability level. The skill, training and education of the migrants and
literacy rate of the receiving country were found as the determinants of
remittances in literature. Buch and Kuckulenz, (2004) established positive
relationship between the illiteracy rates and funds remitted to the home country.
The negative effects of brain drain were counterbalance by the amount of
125
remittances sent to home country from migrant workers (World Bank, 2003).
Rempel and Lobfell, (1978) established weak relation of the remittances toward
the level of skills. The flow of remittances started to decline with the stay period
of the migrants in the host country (Lucas and Stark, 1985). Therefore, the
positive coefficient of education with remittances did not provide any evidence
that the brain drain was a cause of a larger flow of remittances. Rodriguez and
Horton (1994) showed that the education of the migrant worker has no impact on
the flow of remittances in the case of the Philippines. Faini (2006) recommended
that, in general, more skilled migrates were connected with a lesser flow of
remittances.
Finally, the real wage rates in Pakistan affected the inward flow of remittances
negatively and the estimate of the real wage rate was –2.164. These results were
also compared and contrasted with other studies. Swamy, (1981) showed that
the number of migrant workers and their salaries in the host country collectively
brought 90 per cent changes in the amount of remittances into the home country.
Sakka and McNabb, (1999) found out that wage rate available to the migrant in
the host country has a positive and significant impact on the amount of
remittances in case of Egypt.
126
7.3 Error Correction Mechanism
The shock led to a short term departure from the co-integrating equilibrium path
in the real remittances. The model was estimated using the error correction
mechanism. The short run relationship among the variables was captured by
taking the first difference of the variables included in the model. The results of the
error correction mechanism were reported in Table 7.2. The first difference of
real remittances was the dependent variable with the first difference of some
selected explanatory variables in Pakistan. The one year lag of the first
difference of real remittances was also included in the explanatory variables
along with other selected independent variables. This regression analysis
included a sample of 31 observations from 1975 to 2005. The information of two
observations was lost due to inclusion of first difference of variables and lag of
the dependent variable. The estimates of the regression were found significant
on the basis of t statistics.
The results in Table 7.2 revealed that the real remittances were positively related
with the real growth rate and unemployment rate in Pakistan in the short run
period. The estimated coefficients were 0.234 and 0.563 and were significant.
The real remittances were negatively related with literacy rate and spread rates
of the banks in Pakistan. The estimated coefficients were -4.346 and 0.216 and
were significant.
127
The et was the error correction term in this regression model and was calculated
from the long run relationship of the real remittances with the selected
independent variables in Pakistan. The parameter of error correction term was
called adjustment parameter.
Table 7.2: Regression Results relating First Difference of Real Remittances with First Difference of independent variables in Pakistan Variable (Ln) Coefficient Std. Error t-Statistic Prob.
C 0.130 0.069 1.894* 0.0703
∆RREM (-1) 0.494 0.129 3.828*** 0.0008
∆SP -0.216 0.068 -3.167*** 0.004
∆GR 0.234 0.075 3.129** 0.005
∆LR -4.346 2.139 -2.031*** 0.053
∆UR 0.563 0.219 2.568*** 0.017
et (-1) -0.944 0.161 -5.835*** 0.000
R2 0.66
Adjusted R2 0.58
Durbin-Watson 1.72
* Indicated that the coefficient was significantly different from zero at 0.10 probability level. ** Indicated that the coefficient was significantly different from zero at 0.05 probability level. *** Indicated that the coefficient was significantly different from zero at 0.01 probability level. Ln indicated that variables were in the logarithmic form.
128
The value of adjustment parameter was found to be –0.94. The negative sign of
this parameter indicated that level of the real remittances in the short run was
above the level of real remittances in the long run. Therefore, the level of real
remittances showed downward trend and the volume of real remittances was
corrected downward in the short run period approaching to its long run
equilibrium.
7.4 Long Run and Short Run Elasticities
The elasticities of the real remittances with respect to its selected determinants
were considered very important from the point view of policy implications. The
results reported in Table 7.1 and Table 7.2 was useful for estimating the long run
and short run elasticities. The estimates in Table 7.1 and Table 7.2 were
transformed into equation 7.1 and 7.2.
Log RREM = -9.64 + 0.72 Log UR - 2.16 Log RWR - 0.43 Log SP + 0.24 Log GR
- 6.58 Log LR + 3.82 Log RGDP (7.1)
∆ Log RREM = 0.13 + 0.49∆ Log RREM (-1) - 0.21∆ Log SP + 0.23∆ Log GR -
4.34∆ Log LR + 0.56∆ Log UR - 0.94 et (-1) (7.2)
The long run and short run elasticities of the real remittances were calculated by
differentiating equation (7.1) and (7.2). The sign ∆ in equation (7.2) denoted the
short run relationship among the variables. The results of these elasticities were
129
reported in Table 7.3. The real remittances were considered very sensitive to
determinants having elasticity greater than one. For example, in this study, the
elasticity of real remittances with respect to literacy was very high and explained
that the literacy has long-term and short-term negative effect on the real
remittances. The estimated elasticities of real remittances in Pakistan with
respect to unemployment rate were 0.718 and 0.563 in the long run and the short
run respectively in the present study. Schrooten, (2005) estimated long run and
short run elasticities of remittances with respect to the unemployment rate in the
home country and were .3 and 0.29 respectively. The results were found
consistent with the present study.
Table 7.3: Long Run and Short Run Elasticities of Real Remittances in Pakistan
Variables Long Run Elasticities Short Run Elasticities
UR 0.718*** 0.563***
RWR -2.164*** Nil
SP -0.430*** -0.216***
GR 0.236* 0.234***
LR -6.576*** -4.346***
RGDP 3.818*** Nil
* Indicated that the coefficient was significantly different from zero at 0.10 probability level. ** Indicated that the coefficient was significantly different from zero at 0.05 probability level. *** Indicated that the coefficient was significantly different from zero at 0.01 probability level.
130
The estimated elasticity of real remittances in Pakistan with respect to real GDP
was 3.818 significantly different from zero at 0.01 probability level in the long run
period. However, Sakka and McNabb, (1999) found that the elasticity of
remittances with respect to home country GDP was positive in the long run but
reported negative in the short run equal to –0.8. The estimated elasticities of real
remittances with respect to spread rates of banks in Pakistan were –0.430 and –
0.216 in the long run and short run periods. Schrooten, (2005) calculated the
elasticity of spread rate in socialist countries and was reported to be 0.19.
7.5 Elementary Analysis of Workers’ Remittances in Pakistan
It was useful to analyze the remittances data during the period 1973 to 2005 to
know the trend of remittances in Pakistan after the empirical investigation of the
Economic Determinants of International Remittances in Pakistan. Therefore, in
this section, the trend of the international remittances in Pakistan and country
wise remittances has been described. The remittances as a percentage of GDP,
Imports, Exports and Savings in the country have also been calculated to know
the importance of international remittances in Pakistan. The results were
provided in Appendix Table A1, Appendix Table A2 and Appendix Table A3.
131
7.5.1 Trend and Country Wise Remittances in Pakistan
The trend and country wise remittances ($ Million) were presented in Appendix
Table A1 while Appendix Table A2 provide information about the Country Wise
Percentage Share of Remittances in Pakistan. It was revealed that remittances
from the United States were $10 million in 1973, $61.5 million in 1980 and were
$146 million in 1997. It meant that the percentage share of remittances from USA
remained around 3 percent to 10 percent between the periods 1973 to 1997. This
share of remittances from the USA reached from 11 percent to 33 percent during
the period from 1998 to 2005. Therefore, USA emerged as the single largest
source of official remittances in Pakistan during 2003 showing 32.6 percent share
of total remittances. The annual average share of the United Kingdom in total
remittances declined continuously from 53 percent to 5.2 per cent during the
period from 1973 to 1984. Similarly, remittances from Saudi Arabia declined from
the peak level of 52 per cent in the 1984 to 13.7 per cent in 2003. Remittances
from United Arab Emirates increased from 10.4 percent in the 1975 to 19.8
percent in 2003.
The aggregate level of remittances was $136 million in 1972-73, $139 million in
1973-74 and was $1747 million in 1979-80 in Pakistan. The amount of
remittances in Pakistan reached to $2.88 billion in 1982-83 with continuous
increasing pattern. The pattern of remittances in Pakistan was fluctuating after
the period of 1982-83. The volume of remittances has declined from $1.84 billion
132
in 1990-91 to $1.40 billion in 1996-97. However, the remittances again slightly
rose to $1.48 billion during 1997-98 and fell to $0.98 billion in 1999-2000. In the
period of 2002-03, the volume of remittances touched its maximum amount to
$4.23 billion and again came down to $3.87 billion in 2003-04. The volume of
remittances was observed $4.17 billion in the period 2005.
7.5.2 Importance of International Remittances in Pakistan
Pakistan has been one of the major labour exporting countries to the Middle
East. These workers sent a significant amount of their earnings to Pakistan. The
results of remittances as a percentage of GDP, Imports, Exports and Savings in
the country were provided in Appendix Table A3. These results showed that in
the 1973, the international official recorded remittances were equivalent to 2.2
per cent of GDP, 16.9 per cent of exports, 17.2 per cent of imports and were 18.7
per cent of national savings in the country. The remittances were equivalent to
1.7 per cent of GDP, 13.5 per cent of exports, 10.2 per cent of imports and were
22.3 per cent of national savings in 1974. It was observed that the level of
remittances were reached in the country during 1980 as equivalent to 8.2 per
cent of GDP, 73.9 per cent of exports, 36.9 per cent of imports and 57.6 per cent
of national savings. Trend in remittances was reversed during the 1990s. The
remittances declined tremendously to 5.5 per cent of GDP, 39.2 per cent of
exports, 28.0 per cent of imports and 34.3 per cent of national savings in the
1990s. The decline in oil prices, slowing down in economic activities in the major
133
labour importing countries, in particular in the Middle East, together with
increased competition with other labour exporting countries, and freezing of
foreign currency accounts led to a decline in workers’ remittances during the
1990s.
7.6 Impulse Response Function of Remittances Model
The Impulse Response Function (IRF) traced out the response of the dependent
variable in the VAR system to shocks in the error terms. This shock changed the
dependent variable in the current as well as future periods. Therefore, the
Impulse Response Function (IRF) traced out the impact of such shocks on the
variables included in the relevant model for several periods in the future. The
results of generalized Impulse Response Functions (IRF) for the real remittances
models were given in table 7.4 and figure 7.1. The results showed that one
standard error shock to real remittances caused initial response of 32 percent
after first year till it returned to long run equilibrium. One standard error shock to
the real remittances showed an initial response of -0.002 percent in RGDP that
continued slightly rising to 0.02 percent. In the same way, one standard error
shock to the real remittances regarding growth rate attained long run equilibrium
after first year, unemployment rate received long run equilibrium in the first year,
literacy rate got equilibrium after third year with trend and furthermore real wage
rate and spread rate were in the stable long run equilibrium quite in the first year.
134
Table 7.4: Generalized Impulse Response(s) to one S.E. shock in the Equation for LRREM
Horizon
LRREM
LRGDP LGR LUR LLR LRWR LSP
0 .2884 -.0019 -.06312 .0040 -.0023 -.0050 -.0828
1 .3221 .0011 .1009 .0213 -.0001 -.0110 -.1070
2 .3231 .0043 .08575 .0228 .0014 -.0110 -.1044
3 .3270 .0073 .0863 .0256 .0029 -.0115 -.1042
4 .3306 .0104 .0855 .0283 .0045 -.0121 -.1038
5 .3342 .0135 .0849 .0311 .0060 -.0126 -.1033
6 .3379 .0166 .0842 .0339 .0076 -.0131 -.1029
7 .3416 .0196 .0835 .0366 .0092 -.0136 -.1025
8 .3452 .0227 .0829 .0393 .0107 -.0141 -.1020
9 .3488 .0258 .0822 .0421 .0122 .0146 -.1016
10 .3525 .0289 .0816 .0449 .0138 -.0151 -.1012
135
Figure 7.1: Generalized Impulse Response(s) For Remittances Model
136
Chapter 8
Economic Determinants of International Migration
8.1 Introduction
The research studies have investigated that the number of migrant workers
(Migration) was the important determinants of the international remittances. The
remittances were positively related with number of migrants (Russell, 1986), so it
was one of the determinants of remittances. The real earnings of workers and
total number of migrants in the host country were consistently found to have a
significant and positive effect on the flow of remittances (Swamy, 1981).
The determinants of international migration were explored in this chapter
because the causality ran both ways. The remittances were repeatedly found
positively related with migration. Therefore, the estimates of the migration were
considered very important for the policy maker to enhance the process of
migration. A comprehensive and sound migration policy could be formulated on
the basis of the estimated result to fetch more remittances in the country.
Therefore, it was imperative to study the determinants of international migration
with remittances.
137
8.2 Long Run Relationship
The variables in the migration model were found co-integrated exhibiting a long
run relationship. Their long run relationship was determined with ordinary least
square (OLS) method. The results of the OLS estimates were reported in Table
8.1.The estimates were significant on the basis of t statistics. No serial
correlation among the residuals was detected on the basis of D-W statistics. The
results explained that the migration was positively related with the unemployment
rate in the country. The coefficient of unemployment rate was 0.709 and was
highly significant at 0.01 probability level. It showed that when there was
unemployment in the country, people were compelled to leave the home country
for their livelihood. Resultantly they move to foreign countries at very high
psychic and unbearable logistic costs.
The migration was negatively related with real wage rate in the home country.
The estimated coefficient was -1.58 and was highly significant. The low wage
rate forced the workers to move out and caused a massive brain drain from
country especially the skilled and professional workers. The recent initiative of
the Higher Education Commission to provide salary package commensurate with
international level, attracted qualified people to teach in the home country.
The results of the study were compared and contrasted with other studies. Walsh
(1974) estimated the determinants of migration in the Ireland and concluded that
138
the wage rate and unemployment rate differentials between the home and host
countries were significant determinants of the net migration from Ireland. The
results were comparable with the present study.
The migration was found positively related with inflation rate in country and was
highly significant at 0.01 probability level. Therefore, people started to migrate
abroad for fulfilling their basic needs with earned remittances. Particularly, the
worker from the low income segments of the society migrated to foreign countries
under such circumstances.
Table 8.1: Regression Results relating Migration with Independent Variables in Pakistan Variable (Ln) Coefficient Std. Error t-Statistic Prob.
C 9.444 2.094 4.509*** 0.0001
INF 0.453 0.143 3.167*** 0.0037
RREM 0.562 0.111 5.052*** 0.0000
RWR -1.580 0.485 -3.255*** 0.0030
UR 0.709 0.153 4.616*** 0.0001
R2 0.80
Adjusted R2 0.77
Durbin-Watson 1.82
* Indicated that the coefficient was significantly different from zero at 0.10 probability level. ** Indicated that the coefficient was significantly different from zero at 0.05 probability level. *** Indicated that the coefficient was significantly different from zero at 0.01 probability level. Ln indicated that variables were in the logarithmic form.
139
The migration was positively related with the size of the real remittances received
in the country. The estimated coefficient was 0.562 and was highly significant. It
indicated that the increased volume of remittances due to favorable economic
conditions in the host countries provided incentives for the migration.
8.3 Error Correction Mechanism
The short run relationship among the variables in the migration model was
estimated using the error correction mechanism. The short run relationship
among the variables was captured by taking the first difference of the variables
included in the model. The results of the error correction mechanism were
reported in Table 8.2. The first difference of number of migrated workers i.e.
migration was the dependent variable with the first difference of some selected
explanatory variables in Pakistan. The results in Table 8.2 indicated that
migration was positively related with inflation rate, unemployment rate and real
remittances received in the country in the short run. The negative relationship
was observed between migration and real wage rate of the workers in the
country.
The adjustment coefficient was -0.54 showing the dynamic process of long run.
The negative sign of this parameter indicated that level of the migration in the
short run was above the level of migration in the long run.
140
Table 8.2: Regression Results relating First Difference of Migration with First Difference of Independent Variables in Pakistan Variable (Ln) Coefficient Std. Error t-Statistic Prob.
C -0.010 0.047 -0.226 0.822
∆INF 0.272 0.114 2.380** 0.024
∆RREM 0.442 0.182 2.423** 0.022
∆RWR -0.717 0.497 -1.441 0.161
∆UR 1.283 0.366 3.505*** 0.001
Et (-1) -0.544 0.175 -3.095*** 0.004
R2 0.76
Adjusted R2 0.73
Durbin-Watson 1.75
* Indicated that the coefficient was significantly different from zero at 0.10 probability level. ** Indicated that the coefficient was significantly different from zero at 0.05 probability level. *** Indicated that the coefficient was significantly different from zero at 0.01 probability level. Ln indicated that variables were in the logarithmic form.
Therefore, the level of migration showed downward trend and this level of
migration was corrected downward in the short run period approaching to its long
run equilibrium.
141
8.4 Long Run and Short Run Elasticities
The elasticities of migration with respect to selected determinants were
considered important for policy implications. The results reported in Table 8.1
and Table 8.2 was useful for estimating the long run and short run elasticities of
migration. The estimates in Table 8.1 and Table 8.2 were transformed into
equation (8.1) and (8.2).
Log NW = 9.44 + 0.453 Log INF +0.562 Log RREM – 1.580 Log RWR + 0.709
Log UR (8.1)
∆ Log NW = -0.010 + 0.272 ∆ Log INF + 0.442 ∆ Log RREM -0.717 ∆ Log RWR
+ 1.283 ∆ Log UR- 0.544 et (-1) (8.2)
The long run and short run elasticities of migration were calculated by
differentiating equation (8.1) and (8.2). The sign ∆ in equation (8.2) denoted the
short run relationship among the variables.
The results of these elasticities were reported in Table 8.3. The process of
migration was very sensitive to determinants having elasticity greater than one.
For example, in this study, the elasticity of migration with respect to wage rate in
the long run was greater than one and showing that people will migrate with
extended gestation period, however the elasticity of unemployment was greater
142
than one in the short run and people tend to migrate abroad immediately due to
unemployment in the country.
Table 8.3: Long Run and Short Run Elasticities of Migration in Pakistan Variables Long Run Elasticities Short Run Elasticities
INF 0.453** 0.272***
RREM 0.562** 0.442***
RWR -1.580 -0.717***
UR 0.709*** 1.283***
* Indicated that the coefficient was significantly different from zero at 0.10 probability level. ** Indicated that the coefficient was significantly different from zero at 0.05 probability level. *** Indicated that the coefficient was significantly different from zero at 0.01 probability level.
8.5 Impulse Response Function of Migration Model
The results of generalized Impulse Response Functions (IRF) for the migration
model were given in table 8.4 and figure 8.1. The results showed that one
standard error shock to the migration caused initial response of 30 percent after
first year till it returns to long run equilibrium. One standard error shock to the
migration showed an initial response of 0.04 percent in real remittances that
continued slightly rising to.01 percent in the seven year and became smooth after
the seven year. In the same way, one standard error shock to the migration
regarding real wage rate and inflation attained long run equilibrium after second
year with slightly rising trend.
143
Table 8.4: Generalized Impulse Response(s) to one S.E. shock in the Equation for LNW Horizon
LNW LRREM LUR LRWR LINF
0 .2849 .0163 .0513 -.0024 -.0145
1 .3019 .0400 .0571 -.0042 -.0269
2 .3135 .0606 .0616 -.0044 -.0402
3 .3208 .0763 .0649 -.0039 -.0517
4 .3251 .0874 .0670 -.0032 -.0604
5 .3274 .0946 .0683 -.0025 -.0666
6 .3284 .0989 .0691 -.0020 -.0706
7 .3287 .1014 .0695 -.0016 -.0729
8 .3286 .1025 .0696 -.0013 -.0743
9 .3285 .1030 .0697 -.0011 -.0749
10 .3282 .1031 .0696 -.0010 -.0752
144
Figure 8.1: Generalized Impulse Response(s) for Migration Model
145
Chapter 9
Conclusion and Policy Recommendations
It was observed that international remittances were increased over time in
Pakistan with fluctuating pattern. However, the amount of remittances in Pakistan
was found more fluctuating during the period’s from1983-84 to 2004-05. Pakistan
received the volume of remittances $136 million in 1972-73, $139 million in 1973-
74. The amount of remittances in Pakistan reached to $2886 million in 1982-83
with continuous increasing pattern. The volume of remittances was declined from
$1,848.3 million in 1990-91 to $1,409.5 million in 1996-97. However, the
remittances again slightly rose to $1,489.5 million during 1997-98 and fell to $983
million in 1999-2000. In the period of 2002-03, the volume of remittances
reached to maximum level amounted to $4237 million and again came back to
$3872 during 2003-04.
Pakistan has been receiving international remittances from the different countries
of the world because Pakistan has been one of the major labour exporting
countries to the Middle East and other countries of the world. These workers sent
a significant amount of their earnings to Pakistan. It was revealed that
remittances from the United States were $10 million in 1973, $61.5 million in
1980 and were $146 million in 1997. It meant that the percentage share of
remittances from USA remained around 3 percent to 10 percent between the
periods 1973 to 1997. This share of remittances from the USA reached from 11
146
percent to 33 percent during the period from 1998 to 2005. Therefore, USA
emerged as the single largest source of official remittances in Pakistan during
2003 showing 32.6 percent share of total remittances. The annual average share
of the United Kingdom in total remittances declined continuously from 53 percent
to 5.2 per cent during the period from 1973 to 1984. Similarly, remittances from
Saudi Arabia declined from the peak level of 52 per cent in the 1984 to 13.7 per
cent in 2003. Remittances from United Arab Emirates increased from 10.4
percent in the 1975 to 19.8 percent in 2003.
The results of remittances as a percentage of GDP, Imports, Exports and
Savings in the country explained the importance of remittances over the time in
Pakistan. These results showed that in 1973, the international official recorded
remittances were equivalent to 2.2 per cent of GDP, 16.9 per cent of exports,
17.2 per cent of imports and were 18.7 per cent of national savings in the
country. The remittances were equivalent to 1.7 per cent of GDP, 13.5 per cent
of exports, 10.2 per cent of imports and were 22.3 per cent of national savings in
1974. It was observed that the level of remittances were reached in the country
during 1980 as equivalent to 8.2 per cent of GDP, 73.9 per cent of exports, 36.9
per cent of imports and 57.6 per cent of national savings. Trend in remittances
was reversed during the 1990s. The remittances declined tremendously to 5.5
per cent of GDP, 39.2 per cent of exports, 28.0 per cent of imports and 34.3 per
cent of national savings in the 1990s.
147
The remittances were a permanent source of foreign exchanges in the country
and provided supports to the balance of payment. The remittances were used in
the country for different purposes at individual and country level. The households
have utilized their remittances for consumption and for the expansion of their
business. It was the fact that the utilization of remittances was biased toward
consumption rather than investment. It was due to wide spread of poverty and
want of investment opportunities in the country. Therefore, it was necessary to
provide investment friendly atmosphere in the country to enhance investment for
earning more foreign exchange.
The present study examined the effect of various macroeconomic determinants
on the flows of remittances in Pakistan. The home country variables at macro
level were used to obtain determinants of international remittances in Pakistan.
The economic determinants of international migration in Pakistan were also
explored in the study. Two econometrics models were specified to estimate the
determinants of international remittances and migration. The time series data
were collected from secondary sources (i) Federal Bureau of Statistics,
Islamabad; (ii) State Bank of Pakistan, Karachi; and (iii) Bureau of Emigration
and Overseas Employment. The unit root test was applied to know the
stationaitry of the time series data. Most of the time series data were not
stationary. All the data series were stationary after differencing once each series.
Thus, it was concluded that all series used in this research study were of the
same order i.e. I (1). Two approaches were utilized to know the co-integration
148
among the included variables in the remittances and migration models. These
approaches were the augmented Dickey-Fuller (ADF) residual-based test and
the Johansen’s Full Information Maximum Likelihood (FIML) approach. The
variables used in the research were found co-integrated. The long run
relationship was determined among the variables in both models with the help of
ordinary least square method because the variables were co-integrated.
However, error correction model (ECM) was used to observe the short run
relationship and adjustment mechanism toward the long run.
The results of the study showed that the real remittances were positively related
with the unemployment rate as the coefficient of unemployment rate was
estimated 0.719 and was highly significant. The real remittances were also
positively related with the real growth rate and real GDP of the Pakistan. The
estimated coefficients of real growth rate and real GDP were 0.236 and 3.818
and were found significant. The migrant workers sent more remittances under the
circumstances of healthy economic activities for investment motives, which
support the balance of payment in the country. Real growth rate and GDP were
the indicators of the healthy economy. The coefficient of spread rate was
significant with its value –0.43. The coefficient of literacy rate was –6.576 and
was also found significant. Finally, the real wage rates in Pakistan affected the
inflow of remittances negatively and the co efficient of the real wage rate was –
2.164 and was significant.
149
The determinants of migration from Pakistan were also estimated in the study
and results showed that the migration was positively related with the
unemployment rate in the country. The coefficient of unemployment rate was
0.709 and was highly significant. The real wage rates in the home country have
negative impact upon migration with the estimated coefficient of -1.58 and this co
efficient was highly significant. Migration was found positively related with
inflation rate in the country and was highly significant at 0.01 probability level.
The migration was also positively related with the size of the real remittances
received in the country. The estimated coefficient was 0.562 and was highly
significant. The findings of the study were comparable with previous studies.
The government policies are considered very important to accelerate the
migration from Pakistan to the rest of world to fetch more remittances. The
policies those are labour friendly can encourage more remittances in the country.
The bulk of remittances can be used for longer-term growth and income security
in Pakistan. The remittances are private transfers; therefore, it is desirable that
the policy measures adopted for the promotion of remittances must be incentive-
oriented. The following measures were suggested for long run promotion of
remittances in Pakistan.
(i) Pakistan is over populated country having shortage of job vacancies.
The niche markets for labour must be searched through proactive
150
labour policies. This will reduce unemployment and increase the
amount of remittances in the country.
(ii) The saving rate was very low in Pakistan due to wide spread of poverty
in the country. The saving can be mobilized through the foreign
remittances in the country. Therefore, Government must streamline the
saving schemes in country to attract the foreign remittances.
(iii) The policies can be formulated keeping in view the creation of
investment climate and provide incentives to the migrants to promote
domestic commerce in Pakistan. Therefore, policies include facilities to
the migrants for importing machinery and equipment at concession
duty rate for investment in forms of enterprises. Pakistani economic
workers must be provided investment opportunities in export
processing zones.
(iv) The guidance should be provided to the migrants for release of
information by the Overseas Pakistani Foundation (OPF) on the
available credit facilities, savings schemes and business advisory
services. Business counseling should also be provided to the migrant
workers for establishing their business in productive sectors.
(v) The effectiveness of incentive schemes in promoting remittances
depends crucially on the ability of the existing banking network to
compete with the informal market. There was a need to realign the
exchange rate with competitive foreign currencies. It is more realistic to
151
introduce financial innovations to capture migrant remittances as
deposits and channel them to existing small and micro-enterprises.
(vi) The effective strategy for attracting remittances into the formal banking
system should include attempts to expand the branch network further,
in order to effectively link overseas workers with the remittance
receiving families and to take on more of the desirable features of the
services offered by the informal network.
(vii) The elasticity of migration with respect to real wage rate was found
negative and was greater than one. The brain drain in the country
might be controlled by higher salary offers to the skilled persons in the
country.
The research field of remittances is vast and various topics about remittances
may further be questioned, including: (i) The role of micro financial institutions in
linking informal remittances to development of the country can be explored. (ii)
The migrants’ savings can be pooled to form mutual or pension funds for the
welfare of the society (iii) the efficiency of formal and informal channels used for
the remittances can be compared and improvements may be suggested in the
existing banking system in its ability to compete effectively with informal
channels. (iv) The impact of remittances upon the different macroeconomic
variables can be estimated to know the further importance of the remittances in
the country. (v) The role of government policies about the incentive schemes to
channel remittances in productive uses can be assessed. (vi) The different
152
prospects of brain drain in Pakistan due to migration can be highlighted and their
solution can be presented.
These questions are considered important for the future avenues and may
provide some help to the researchers working relating to international migration
and remittances in future. Therefore, further issues can be explored on the same
topic in the light of the results of present research study.
153
References
Amjad, R., (1986), “Impact of Worker’s Remittances from the Middle East on Pakistan’s Economy”, Paper presented to Third Annual General Meeting of the Pakistan Society of Development Economics, Islamabad
Alburo, F.A. and D.I. Abella, (1992),"The impact of informal Remittances of Overseas Contract Workers' Earnings on the Philippine Economy" New Delhi, ARTEP (mimeo) Alberto Rey Poveda (2007) “Determinants and consequences of internal and international migration: The case of rural populations in the south of Veracruz, Mexico” Demographic Research: Volume 16, Article 10 http://www.demographic-research.org 287 A. Castaldo and B. Reilly (2007) “Do Migrant Remittances affect the Consumption Patterns of Albanian Household?” South-Eastern Europe Journal of Economics 1 (2007) 25-26 54 Agarwal, R. and A.W. Horowitz, (2002), “Are International Remittances Altruism or Insurance? Evidence from Guyana Using Multiple-Migrant Households”, World Development, Vol. 30, pp. 2033-2044 Aydas, O. T., Neyapti, B. and K. M. Ozcan, (2004), “Determinants of Workers Remittances: The Case of Turkey”, Department of Economics, Bilkent University, Ankara, Turkey, Discussion Paper Burney, N., (1988), “A Macroeconomic Analysis of the Impact of Worker’s Remittances from the Middle East on Pakistan’s Economy” ILO, (ARTEP), Working Paper No.8 Benjamin Davis and Guy Stecklov (2002) “Domestic and international migration from rural Mexico: Disaggregating the effects of network structure and composition” Population Studies, 56 (2002), 291-309. Benoit, K. and A. Vencatachellum, (2002), “An Empirical Analysis of Compulsory and Voluntary Remittances among Domestic Workers in Tunisia” Institute of Applied Economics, HEC Montreal Buch, C. M., and A. Kuckulenz, (2002), “Worker remittances and capital flow” Kiel Institute for World Economic, Working Paper Series, No. 1130 Bauer T. and I. Gang (1998), “Temporary migrants from Egypt: how long do stay abroad?” Institute for the Study of Labour, Bonn University, D.P. No. 3
154
Bhagwati J. and K. Hamada (1974), “The brain, international integration of markets for professionals and unemployment: a theoretical analysis”, Journal of Development Economics, Vol. 1 Benhabib, J. (1996), “On the political economy of immigration”, European Economic Review, Vol. 40 pp1737-1743 Borjas, G. and M. Miller (2003) “The Age of Migration” Basingstoke: Palgrave Borjas, G. (2001) “Heaven’s door: Immigration policy and the American economy”. Princeton University Press Borjas G. (1989), “Immigrant and emigrant earnings: a longitudinal study”, Economic Inquiry, Vol. 27, pp 21-37 Batzlen, C., (2000), “Remittances and Investments in South Asia: A Case Study of Pakistan”, Development Economics and policy, New York, Oxford, Vol. 21, pp 332 Castles, Stephen and M.J. Miller, (2003) “The Age of Migration New York and London” The Guilford Press. Chiswick, B. R. and T.J. Hatton, (2002), “International migration and the integration of labour markets In Globalization in Historical Perspective” The University of Chicago, Chicago Press Chami, R., Fullenkamp and S. Jahjah, (2003) “Are Immigrant Remittance Flows a source of capital for Development?” IMF Working Paper, No. 03/189 Chandavarkar, A. G., (1980), “Use of Migrants Remittances in Labor-Exporting Countries,” Finance and Development Vol. 17, pp 36-39 Docquier, F. and A. Marfouk, (2004),”Measuring the international mobility of skilled workers”, Policy Research Working Paper No. 3382, World Bank Datta, P. (1998), “Migration to India with special reference to Nepali migration”, unpublished doctoral dissertation, University of Calcutta, Kolkata, and West Bengal, India. Datta, P. (2002), “Nepali migration to India” Paper presented in the Regional Population Conference, South East Asia’s Population in a changing Asian Context organized by International Union for The Scientific Study of Population. Bangkok, Thailand.
Davidson, J. E. H., Handry, D. F Srba, F. and S. Yeo, (1978), “Econometric Modeling of the Aggregate Time Series Relationship between consumers
155
expenditures and income in the United Kingdom”, Economic Journal, Vol. 88, pp 661-692 Dickey, D. A., and S., Pantula, (1987), “Determining the order of the differencing in autoregressive processes”, Journal of Business and Economics Statistics, Vol. 15, pp 455–461
Dickey, D. A., Bell, W. R., and Miller, R. B. (1986), “Unit Roots in Time Series Models: tests and implications”, The American Statistician, Vol. 40, pp 12-26
Engle, R. F. and Granger, C. W. J. (1987), “Co-integration and Error correction: Representation, Estimation and Testing”, Econometrica, Vol. 55, pp 251-276 Elbadawi, I. A., and R.R. Rocha, (1992), “Determinants of Expatriate Workers’ Remittances in North Africa and Europe”, World Bank Working Paper Series, No. 1038- 20 Elsakka, M.I.T., and R. Mcnabb, (1999), “The Macroeconomic Determinants of Emigrant Remittances”, World Development, Vol. 27, pp.1493-1502 Facchini, Giovanni and G. Willman, (2001),“The Political Economy of International Factor Mobility”, SIEPER Working Paper 00-20, Stanford University Frank, Robert (2001), "Philippine Town Plies a Road to Riches via Monthly Stipends: But are Pozorrubians turning into Slakers as Relatives Abroad Send Home Cash?" Wall Street Journal Faini, R., (1994), “Worker Remittances and the Real Exchange Rate: A Quantantive Frame Work”, Journal of Population Economics, Vol. 7, pp 235-245 Faini R. (2005b), “Migration, remittances and Poverty”, International Migration and Asylum, Palgrave, Mac Millan for WIDER Faini, R. (2006), “Remittances and the brain drain”, CEPR Discussion Paper No. 4031, CEPR London. Filiz Garip (2006) “Social and Economic Determinants of Migration and Remittances: An Analysis of 22 Thai Villages” Department of Sociology, Princeton University Gubert, F., (2002), “Do Migrants Insure Those Who Stay Behind? Evidence from the Kayes Area (Western Mali)”, Oxford Development Studies, Vol. 30, pp. 267-87 Govt. of Pakistan (2006), “Fifty year Statistics of Pakistan”, Federal Bureau of Statistics, Islamabad
156
Govt. of Pakistan, “Statistical Bulletin, Different Issues”, State Bank of Pakistan, Karachi Govt. of Pakistan (2005), “Handbook of Statistics on Pakistan Economy”, State Bank of Pakistan, Karachi Govt. of Pakistan, “Economic Survey of Pakistan (Various Issues)” Finance Division, Islamabad Govt. of Pakistan, Bureau of Emigration and Overseas Employment, Islamabad Glytsos, N., (1988), “Remittances in Temporary Migration: A Theoretical Mode and its Testing with the Greek-German Experience,” Weltwirtschafliches Archiv 124: 524-548. Glytsos, N. and L. Katselli, (1986), “Theoretical and Empirical Determinants of International Labor Mobility: A Greek-German Perspective,” Center for Economic Policy Research, Discussion Paper Series No.148.
Granger, C. W. J. (1981), “Some properties of time series data and their use in Econometric Model Specification”, Journal of Econometric, Vol. 16, No.1, pp 121-130
Granger, C. W. J., (1983), “Co-integrated variables and Error Correction Models”, UCSD, Discussion Paper
Granger, C. W. J. and P. Newbold, (1974), “Spurious Regression in Econometrics”, Journal of Econometrics, Vol. 2, pp 111-120 Grossman, Gene M. and E., Helpman (1994), “Protection for Sale”, The American Economic Review, Vol. 84, pp. 833-50. Hoddinott, J., (1992), “Modelling Remittance Flows in Kenya,” Journal of African Economics, Vol. 1, No. 1, pp 233-270. Hyder Zulfiqar, (2003), “Workers, Remittances, Resident FCAs and Kerb Premium: A Cointegration Analysis”, State Bank of Pakistan, Working papers, April 2003. Harrison A. et. al. (2004), “Working abroad-the benefits flowing from nationals working in other economies”, OECD, Round Table on Sustainable Development 2004 Harris J. R. and M. P. Todaro, (1970), “Migration, unemployment and development: A two sector analysis” American Economic Review, 60: 126-142.
157
Hamilton, B. and J. Whalley, (1984), “Efficiency and distributional implications of global restrictions on labour mobility”, Journal of Development Economics, Vol. 14, pp 6175 Hatton, T.J. and J.G. Williamson, (1998), “The age of mass migration: Causes and Economic Impact”, Oxford University Press
Hallam, D. and R. Zanoli, (1993), “Error Correction Models and Agricultural Supply Response”, European Review of Agricultural Economics, 20, 151-66
Holden, D. and R. Perman, (1994), “Unit Roots and Cointegration for the Economists”, St. Martin s’ press, New York Herbert Braucker and Philipp J.H. Schraoder (2005), “International Migration with HeterogeneousAgents: Theory and Evidence” Aarhus School of Business, DIW Berlin and IZA, Bonn. Ilahi, N. and S. Jafarey, (1999), “Guest worker Migration, Remittances, and the Extended Family: Evidence from Pakistan”, Journal of Development Economics, Vol. 58, 485-512 International Monetary Fund (2005), “World Economic Outlook: Globalization and external Imbalances”, Chapter 2, Washington, DC. Johansen, S., and K. Juselius, (1992), “Testing Structural Hypotheses in a Multivariate Cointegration Analysis of the PPP and UIP for UK”, Journal of Econometrics, Vol. 53, pp.211-44
Johnson G., Whitelaw, (1974), “Urban-Rural Income Transfers in Kenya: An Estimated Remittances Function” Economic Development and Cultural Change, Vol. 22, No. 3 pp 473–79
Johansen, S. (1988), “Statistical Analysis of Cointegrating Vectors”, Journal of Economic Dynamics and Control, Vol. 12, pp 231-254
Johansen, S. (1992), “Determination of Cointegrating Rank in the Presence of a Linear Trend”, Oxford Bulletin of Economics and Statistics, Vol. 54, pp 383-397
Johansen, S. and K. Juselius, (1990), “Maximum Likelihood Estimation and Inference on Co-integration – with Application to the Demand for Money”, Oxford Bulletin of Economics and Statistics, 52, 170-209 Juthathip Jongwanich (2007) “Workers’ Remittances, Economic Growth and Poverty in Developing Asia and the Pacific Countries” United Nations Economic and Social Commission for Asia and the Pacific, UNESCAP Working Paper No 07/01
158
Kazi, Shahnaz, (1988), “Domestic Impact of Remittances and Overseas Migration: Pakistan" ILO (ARTEP), Working Paper No.7 Koksal, N. E., and T. Liebig, (2005), “Principal Channels and Costs of Remittances: The Case of Turkey”, Migration, Remittances and Development, OECD, pp. 103-122 Knowles, J. and L. Anker, (1981), “An Analysis of Income Transfers in a Developing Country: The Case of Kenya,” Journal of Development Economics, Vol. 8, pp 205- 226 Lewis A. W. (1954), “Economic development with unlimited supplies of labour” Manchester Lucas, R. and O. Stark, (1985), “Motivations to Remit: Evidence from Botswana,” Journal of Political Economy, Vol. 93, pp 901-918
Luthepohl, H., (1993), “Introduction to Multiple Time Series Analysis”, Second Edition, Springer-Verlag, New York
Massey, D.S., Goldring, L. and Durand, J., (1994), “Continuities in Transnational Migration: an Analysis of Nineteen Mexican Communities”, American Journal of Sociology, pp 11492-1533 Malik, S. J. and Sarwar, N., (1993), “Some Tests for Differences in Consumption Patterns: The Impact of Remittances Using Household Income and Expenditure Survey Data of Pakistan 1987-88”, The Pakistan Development Review, Vol. 2, No. 4 Mayda, A. M., (2007),”International migration: A panel data analysis of the determinants of bilateral flows”, Department of Economics and School of Foreign Service, Georgetown University. CReAM Discussion Paper No 07/07 Milena Jovicic and Radmila Dragutinovic Mitrovic (2006) “Macroeconomic Analysis of Causes and Effects of Remittances: A Panel Model of the SEE Countries and a Case Study of Serbia” Global Development Network Southeast Europe Nil Demet Güngör and Aysıt Tansel (2007) “Brain Drain from Turkey: The Case of Professionals Abroad” Economics Department, Atılım University, Ankara 06836 Turkey
Nickell, S. (1985), “Error Correction, Partial Adjustment and All That: An Expository Note”, Oxford Bulletin of Economics and Statistics, 47 (2), 119-29. Neyapti, B., (2003), “Trends in Workers Remittances: A worldwide overview”, Emerging Markets Finance and Trade, forthcoming.
159
Poirine, Bernard, (1997), “A Theory of Remittances as an Implicit Family Loan Arrangement,” World Development, Vol. 25, pp. 589–611 Portes, Alejandro (1999), “Conclusion: towards a new world: the origins and effects of transnational activites”, Ethnic and Racial Studies, Vol. 22, No. 2 Piore, M.J., 1979, “Birds of passage: Migrant labour in industrial societies”, Cambridge: Cambridge University Press.
Perron, P. (1990), “Testing for a Unit Root in a Time Series with a Changing Mean”, Journal of Business and Economic Statistics, Vol. 8, No. 2, pp153-162
Perron, P. and Vogelsang, T. J. (1991), “Testing For a Unit Root in a Time Series with a Changing Mean: Corrections and Extensions”, Mimeo, Department of Economics, Princeton University, Princeton.
Pesaran, H. M. and Pesaran, B. (1997), “Working with Microfit 4.0: Interactive Econometric Analysis”, Oxford University Press.
Phillips, P. C. B. and P. Perron (1988), “Testing for a Unit Root in Time Series Regressions”, Biometrika, Vol. 75, pp 335-346 Rahman, A., (2003), “The Determinants of Foreign Worker Remittances in the Kingdom of Saudi Arabia”, Department of Economics, King Saud University, P. O. Box 2459 Riyadh 11451
Rempel H., Lobdell R. (1978), “The Role of Urban-to-Rural Remittances in Rural Development” Journal of Development Studies Vol. 14, No. 3, pp 324–41
Rodrik, D. (1995), “Political economy of trade policy”, In Grossman, G. and Rogoff, K., editors”, Handbook of International Economics, Vol.3, chapter 28, pages 1457—1494. North-Holland Elsevier Science, The Netherlands. Russell, S. S., (1986), “Remittances from International Migration: A Review in Perspective”, World Development, Vol. 14, No. 6, pp. 677-696. Rotte, R. and V. Michael, (1998) "Determinants of International Migration: Empirical Evidence for Migration from Developing Countries to Germany," IZA Discussion Papers I2, Institute for the Study of Labor Rodriguez E. and S. Horton (1994), “International return migration and remittances in the Philippines”, in D. O’Connor and L. Farsakh (eds.), Development Strategy, Employment and Migration. Country Experiences, OECD Development Centre, Paris
160
Richard H. Adams, Jr. (1996) “Remittances, Income Distribution, and Rural Asset Accumulation” FCND Discussion Paper No. 17, Washington, D.C. International Food Policy Research Institute Solimano, A. (2002), “Globalizing Talent and Human Capital: Implications for Developing Countries”, ABCDE Conference Europe, Oslo
Said, S. E. and Dickey, D. A. (1984), “Testing for Unit Roots in autoregressive-Moving Average Models of Unknown Order”, Biometrika, 71, 599-607.
Sargan, J. D. (1964), “Wages and Prices in the United Kingdom: A Study in Econometric Methodology. Quantitative Economics and Econometric Analysis”, Basil Blackwell, Oxford Stark, O. and D. Bloom, (1985), “The New Economics of Labor Migration” The American Economic Review, Vol. 75, No. 2, pp 173-178. Straubhaar, T., (1986), “The Determinants of Workers’ Remittances: The Case of Turkey,” Weltwirtschafliches Archiv 122: 728-740. Schrooten, M., (2005), “Bringing Home the Money -What Determines Worker's Remittances to Transition Countries?” Discussion Paper Series A No.466 at the Institute of Economic Research Hitotsubashi University Kunitachi, Tokyo, Japan. Schiopu, L., and N. Siegfried (2006), “Determinants of Workers’ Remittances: Evidence from the European Neighboring Region”, Working Paper Series No. 688, European Central Bank Swamy, G., (1981), International Migrant Workers’ Remittances: Issues and Prospects, World Bank, Staff Working Paper, No. 481. Sekaran, U., (1992), “Research Methods for Business”, 2nd Edition, Wiley & Sons Inc., New York Sims, C.A. (1972), “Money, Income and Causality” American Economic Review, Vol 42 pp. 540-552 Shivani Puri and Tineke Ritzema,(1999), “Migrant Worker Remittances, Micro-finance and the Informal Economy: Prospects and Issues” Social Finance Unit, Working paper No. 21, Enterprise and Cooperative Development Department, International Labour Office Geneva. Sjaastad L. A. (1962) “The costs and returns of human migration” Journal of Political Economy, 70: 80-93 Stark, O. (1990), “The migration of labor” Blackwell, Cambridge
161
Stark O. and D. E. Bloom (1985), “The new economics of labor migration” American Economic Review, 75: 173-178 Stark O. and E. Taylor (1989) “Relative deprivation and international migration” Demography, 26 Todaro M. P. (1976) “International migration in developing countries” ILO, Geneva UKIE (2004), Office of the Committee for European Integration of the Republic of Poland Taylor, E. (1994), “International migration and economic development: a micro economy-wide analysis”, Employment and Migration, OECD, Paris Todaro, M. P. A. (1969), “Model of labour migration and urban unemployment in less developed countries” American Economics Review138–148 Wahba, S., (1991), “What Determines Workers’ Remittances”, Finance and Development, Vol. 28, No. 4, pp. 41-44 W.E. Griffiths, R. Carter Hill and G.G. Judge, (2001), “Undergraduate Econometrics”, Second Edition, John Wiley and Sons, Inc World Bank (2003), “Global Development Finance, Chapter 7”, the World Bank, Washington World Bank (2006), “Global Economic Prospects: Economic implications of remittances and migration”, the World Bank, Washington Walsh, B.M. (1974), “Expectations, Information, and Human Migration: Specifying an Econometric Model of Irish Migration to Britain” Journal of regional Science 14,107-120. Yao Lu and Donald J. Treiman (2007) “The Effect of Labor Migration and Remittances on Children’s Education among Blacks in South Africa” California Center for Population Research, UCLA
162
APPENDIX Appendix Table A1: Country Wise Remittances ($ Million) in Pakistan
Years/Countries
Saudi
Arabia UAE UK USA Others Total
1973 7.9 0.0 72.1 10.0 46.0 136.0
1974 10.5 0.0 55.4 14.4 58.7 139.0
1975 17.3 22.6 74.1 19.2 82.9 216.0
1976 46.4 62.4 54.4 25.8 150.1 339.0
1977 158.8 117.8 49.3 29.3 222.5 577.7
1978 464.1 208.3 76.7 51.5 355.7 1156.3
1979 594.4 205.8 119.1 53.6 425.0 1397.9
1980 795.5 216.8 149.7 61.5 523.6 1747.1
1981 984.3 265.4 184.9 71.0 610.3 2115.9
1982 1129.5 224.9 121.3 72.1 677.1 2224.9
1983 1442.0 344.7 161.7 133.5 803.6 2885.5
1984 1441.1 309.0 141.8 105.8 739.8 2737.4
1985 1245.2 301.3 136.0 105.4 658.0 2445.9
1986 1162.9 311.5 223.3 194.5 703.2 2595.3
1987 945.5 278.2 204.9 191.9 658.0 2278.6
1988 821.8 216.6 215.9 177.5 580.9 2012.6
1989 691.6 161.2 171.1 174.8 698.3 1897.0
1990 626.4 143.0 178.2 209.2 785.5 1942.4
Source: Calculated from Handbook of Statistics on Pakistan Economy (2005), SBP
163
Appendix Table A1: Country Wise Remittances ($ Million) in Pakistan (Cont..)
Years/Countries
Saudi
Arabia UAE UK USA Others Total
1991 682.0 172.0 180.1 190.2 624.0 1848.3
1992 516.2 105.1 137.0 150.3 558.9 1467.5
1993 525.9 97.8 114.0 157.8 666.7 1562.2
1994 493.7 99.4 101.2 122.5 628.9 1445.6
1995 554.1 178.3 110.0 141.1 882.7 1866.1
1996 503.2 161.9 109.7 141.9 544.4 1461.2
1997 418.4 164.4 97.9 146.3 582.5 1409.5
1998 474.9 207.7 98.8 166.3 541.9 1489.6
1999 318.5 125.1 73.6 82.0 461.1 1060.2
2000 309.9 147.8 73.3 80.0 372.9 983.7
2001 304.4 190.0 81.4 134.8 375.9 1086.6
2002 376.3 469.5 151.9 779.0 612.3 2389.1
2003 580.8 837.9 273.8 1237.5 1306.9 4236.9
2004 565.3 597.5 333.9 1225.1 1149.8 3871.6
2005 627.2 712.6 371.9 1294.1 1163.1 4168.8
Source: Calculated from Handbook of Statistics on Pakistan Economy (2005), SBP
164
Appendix Table A2: Country Wise Percentage Share of Remittances in Pakistan
Years/Countries
Saudi
Arabia UAE UK USA Others Total
1973 5.8 0.0 53.0 7.3 33.8 100.0
1974 7.6 0.0 39.8 10.4 42.2 100.0
1975 8.0 10.4 34.3 8.9 38.4 100.0
1976 13.7 18.4 16.0 7.6 44.3 100.0
1977 27.5 20.4 8.5 5.1 38.5 100.0
1978 40.1 18.0 6.6 4.5 30.8 100.0
1979 42.5 14.7 8.5 3.8 30.4 100.0
1980 45.5 12.4 8.6 3.5 30.0 100.0
1981 46.5 12.5 8.7 3.4 28.8 100.0
1982 50.8 10.1 5.5 3.2 30.4 100.0
1983 50.0 11.9 5.6 4.6 27.8 100.0
1984 52.6 11.3 5.2 3.9 27.0 100.0
1985 50.9 12.3 5.6 4.3 26.9 100.0
1986 44.8 12.0 8.6 7.5 27.1 100.0
1987 41.5 12.2 9.0 8.4 28.9 100.0
1988 40.8 10.8 10.7 8.8 28.9 100.0
1989 36.5 8.5 9.0 9.2 36.8 100.0
1990 32.2 7.4 9.2 10.8 40.4 100.0
Source: Calculated from Handbook of Statistics on Pakistan Economy (2005), SBP
165
Appendix Table A2: Country Wise Percentage Share of Remittances in Pakistan (Cont..)
Years/Countries
Saudi
Arabia UAE UK USA Others Total
1991 36.9 9.3 9.7 10.3 33.8 100.0
1992 35.2 7.2 9.3 10.2 38.1 100.0
1993 33.7 6.3 7.3 10.1 42.7 100.0
1994 34.1 6.9 7.0 8.5 43.5 100.0
1995 29.7 9.6 5.9 7.6 47.3 100.0
1996 34.4 11.1 7.5 9.7 37.3 100.0
1997 29.7 11.7 6.9 10.4 41.3 100.0
1998 31.9 13.9 6.6 11.2 36.4 100.0
1999 30.0 11.8 6.9 7.7 43.5 100.0
2000 31.5 15.0 7.4 8.1 37.9 100.0
2001 28.0 17.5 7.5 12.4 34.6 100.0
2002 15.8 19.7 6.4 32.6 25.6 100.0
2003 13.7 19.8 6.5 29.2 30.8 100.0
2004 14.6 15.4 8.6 31.6 29.7 100.0
2005 15.0 17.1 8.9 31.0 27.9 100.0
Source: Calculated from Handbook of Statistics on Pakistan Economy (2005), SBP
166
Appendix Table A3: Remittances as Percentage of GDP, Exports, Imports and Savings
Year GDP Exports Imports Savings
1973 2.2 16.9 17.2 18.7
1974 1.7 13.5 10.2 22.3
1975 2.1 20.8 10.2 32.1
1976 2.8 29.8 16.4 25.1
1977 4.2 50.6 24.8 33.6
1978 7.2 88.2 41.2 47.8
1979 7.8 81.8 38.0 62.0
1980 8.2 73.9 36.9 57.6
1981 8.5 71.5 39.1 49.8
1982 8.0 90.3 39.6 50.8
1983 11.2 107.1 53.9 59.2
1984 9.9 98.9 48.2 58.4
1985 8.7 98.2 41.4 60.8
1986 9.0 84.5 46.1 54.7
1987 7.6 61.8 42.4 40.3
1988 5.9 45.2 31.5 38.5
1989 5.3 40.7 27.0 33.6
1990 5.5 39.2 28.0 34.3
Source: Calculated from Handbook of Statistics on Pakistan Economy (2005), SBP
167
Appendix Table A3: Remittances as Percentage of GDP, Exports, Imports and Savings (Cont..)
Year GDP Exports Imports Savings
1991 4.6 30.1 24.3 28.6
1992 3.4 21.3 15.9 17.6
1993 3.4 22.9 15.7 22.3
1994 3.1 21.2 16.9 17.7
1995 3.4 22.9 18.0 21.3
1996 2.5 16.8 12.4 19.6
1997 2.5 16.9 11.9 19.2
1998 2.6 17.3 14.7 16.4
1999 1.8 13.6 11.2 14.4
2000 1.4 11.5 9.5 8.5
2001 1.6 11.8 10.1 9.3
2002 3.6 26.2 23.1 17.9
2003 5.5 38.0 34.7 24.7
2004 4.3 31.4 24.8 21.5
2005 4.0 29.0 20.2 25.0
Source: Calculated from Handbook of Statistics on Pakistan Economy (2005), SBP
168
LIST OF ABBREVIATIONS
ADF Augmented Dickey Fuller
AIC Akaike Information Criterion
ARTEP Asian Regional Team for Employment Promotion
BZU Bahauddin Zakariya University
ECM Error Correction Mechanism
EU European Union
FCA Foreign Currency Account
FCBC Foreign Currency Bearer Certificate
FEBC Foreign Exchange Bearer Certificate
FIML Johansen’s Full Information Maximum Likelihood
GDP Gross Domestic product
GOP Government of Pakistan
ILO International Labor organization
IMF International Monetary Fund
IZA International Zeolite Association
KSA Kingdom of Saudi Arabia
LR Likelihood Ratio
MPC Marginal Propensity to Consume
NWFP North Western Frontier Province
NELM New Economics of labour Migration
OECD Organization for Economic Co operation
and Development
OLS Ordinary Least Square
SBC Schwarz Bayesian Criterion
SBP State Bank of Pakistan
SEE Southeast Europe
S&E Science and Engineering
169
LIST OF ABBREVIATIONS (Cont..)
SPSS Statistical Package for Social Sciences
UK United Kingdom
UOS University of Sargodha
USA United State of America
VAR Vector Auto Regressive