Out-migration in rural Pakistan: does household poverty status matter?
ul Haq, Rizwan ; Jahangeer, Ajmal ; Ahmad, Azkar 等
Apart from a few studies on the internal migratory flows in the
country, there has been a lack of in-depth analysis although a number of
data sets are available on migratory movements in Pakistan. Further
those studies conducted on determinants of migration, incorporated only
current socio-economic and demographic indicators. The data of Pakistan
Panel Household Survey (PPHS) provide us an opportunity to study the
determinants of out-migration taking place between 2002 and 2010.
Further, we assume that decision to migrate is not an individual's
decision rather it is a collective decision of the household, we have
included household level characteristics in the analysis. The results
show that ownership of land is negatively associated with out-migration,
international as well as within country migration. Further, the economic
(poverty) status of the household does not show any significant
association with out-migration.
Keywords: Keywords: Out-migration, Poverty, Rural, Pakistan
1. INTRODUCTION
Movement of the people within the geographical and administrative
boundaries of a country is known as internal migration. Researchers
regard the movement to urban areas from both rural and less-advanced
urban areas as more important, yet studying the dimensions of movement
between rural areas is worth investigating. Scholars assert economic
incentives as the main motive behind the rural-urban movement; various
unforeseeable factors, however, may also stimulate the human flows. In
Pakistan, the phenomenon of internal migration is as old as the
inception of the country as Helbock (1975a) maintained, while studying
life-time migrants in 12 largest cities of the country in 1961, that
almost every 7th person residing in these cities had come from a
different district.
We may split studies on migration into two categories: macro level
studies that focus on flows, magnitude, and future forecast on
migration; and micro level studies that lay emphasis on households or
individuals and how decisions are made to migrate and what is the impact
of such decisions on the well-being of the migrants, and those who are
left behind. In the presence of a dearth of research on migration, we
find most of the studies at the macro level indicators with a very few
at the micro level in Pakistan [e.g. Khan and Shahnaz (2000); Arif
(2005)].
In Pakistan, majority of the macro level studies relied on the
census data that provide a detailed picture of the movement and trends
of migration at the country level. Using the last census conducted in
1998, five studies focused on the migratory movements in the country: at
national level Karim and Nasar (2003) conducted a study on the
inter-district and inter-provincial movements; four other studies
provided a detailed description of inter-provincial movements'
pattern to prepare a socio-economic and demographic profile of the
provinces [Khatak (2003); Chaudhry (2003); Naeem (2003); Rukanuddin and
Chaudhry (2003)].
Apart from the census-based studies, a number of studies have used
survey data to understand various dimensions of internal migration:
Akram et al (2003) undertook a study on the migratory flows using the
1998-1999 Pakistan Integrated Household Survey (PIHS) data on the
province of Punjab; Memon (2005) conducted a district-level study on
migration using the Labor Force Survey (LFS) and (PIHS); Mahmud, et al.
(2010) studied the impact of social sector development on internal
migration using the LFS data; and Hamid (2010) used various LFS rounds
for studying the gender dimension of internal migration.
Some studies focused on the migration pattern at the micro level
using nationally representative data: Khan and Shahnaz (2000) analyzed
the determinants of internal migration in the country using the
1996-1997 LFS data; Arif (2005) used the Paksitan Socio- economc Survey
2001-02, a nationally representative survey conducted by Pakistan
Institute of Development Economics, to study the relationship between
migration and well-being of a household. Others looked at the impact of
migration at micro level using small surveys: Oda (2007) focused on
Chakwal district; number of studies focused the determinants and impact
of migration on migrants and those who were left behind in the district
of Faisalabad [Farah, Zafar and Nawaz (2012); Farooq and Cheema (2005);
Farooq and Javed (2009); Farooq, Mateen and Cheema (2005)].
2. HISTORICAL PERSPECTIVE REGARDING INTERNAL MIGRATION IN PAKISTAN
In Pakistan migration is an old phenomenon, and dates back to the
inception of the country. Soon after the division of India, huge
migration took place from and to India. Extensive research work lacks on
the migration flows within the country, although several data sets exist
providing ample data on directional flows, incidence, income
differentials of migrants and non-migrants, and determinants of
migration.
Census data provides us a detailed picture of the pattern of
inter-district and inter-provincial migration within Pakistan, although
it encounters some shortcomings. Based on 1961 and 1972 census data,
Helbock (1975b), while studying urban population growth in Pakistan,
maintained that population of the country living in larger cities
increase to 50.1 percent in 1972 as compared to 33.4 percent in 1961,
thereby pointing at a rural-urban movement in the country during the
study period. Irfan (1981), while studying the migration trends provided
by Population, Labor Force and Migration (PLM 1979) survey, made a
similar argument that internal migration is becoming rural-urban and
long distance in Pakistan.
The most recent census data (1998) revealed that the volume of
overall migrants has increased whereas the proportion of migrants in
total population has decreased [Karim and Nasar (2004)]. This finding
may be due to the weakness in the criteria of 'defining migrants
used by the 1998 census which did not capture intra-district movement.
Results from various surveys conducted in Pakistan have also come
up with the finding that the prevalent migratory flows in Pakistan are
from rural areas and towards urban areas. PIHS 1998 data suggest that
almost 40 percent of the male migrants are rural to urban followed by
rural to rural movement [Memon (2005)]. Pakistan Socioeconomic Survey
(PSES) data also show the direction of migration from rural to urban and
rural to rural [Arif (2005)]. The PLM 1979, which is indeed quite an old
data, indicated that movement between rural areas was prevalent among
internal migrants in Pakistan. On the other hand, the LFS reveals that
main flow of migrants in Pakistan is between urban areas. This data set,
however, exclude population below 10 years of age while studying
migratory flows.
Among the provinces, urban areas of Sindh were found to be the main
recipient of the internal life time migrants from rural areas of Punjab
and Khyber Phakhtoonkhwa (KP), whereas their counterparts in the
provinces of Balochistan and Sindh move quite less in numbers [Karim and
Nasar (2004)]. The authors argued that from rural areas of Sindh there
might be temporary migrants, which the census data was unable to
capture. Khan and Shahnaz (2000) came up with the same finding using the
multivariate analysis that residents of the province of Punjab are more
likely to migrate.
Using the LFS data, Memon (2005) found that majority of the
migrants are family migrants or those who migrate due to marriage,
whereas only 20 percent of the migrants move due to some economic
reason. Arif (2005) using the PSES data came up with the finding that 61
percent of migrants moved due to economic reason, whereas for female the
prevalent reason is marriage or joining family. Mahreen and Mahmood
(2010) using the latest LFS data suggested that it is mainly the
improvement of economic status which lead people to move as compared to
public utilities.
Surveys focusing on migration generally include a question
regarding the reasons for migration. Compiling various sources of
migration data, Table 3 presents the reported reason of migration. As
may be observed from the table the main reason for migration has been
marriages and family reunion in Pakistan. Combining all the economic
reasons reveal that almost one-fifth of the migrants moved due to some
economic (monetary) incentives.
Considering the gender differentials among the migrants, 1998
census reveals that head of the household's spouses are the leading
long-term migrants followed by the daughters. Among males migrants,
one-third of the migrants are head of household's sons with
one-fifth of the migrants are main bread-earner in the household [Karim
and Nasar (2004)]. Regarding the gender dimension of migration, Hamid
(2010) concluded using data from several rounds of the LFS that female
migrants dominate all four types of moves; especially they outnumber
their male counterparts in the urban-urban migration in the country.
Almost half of the female migrants moved due to their marriages.
The majority of the male migrants moving towards urban areas have
at least 6 years of schooling according to 1996-97 LFS survey [Khan and
Shahnaz (2000)]. On the other hand majority of the people who move
between rural areas have no formal education. Among female migrants,
majority of those who move have no formal education. The pattern may be
characterized as better educated people moving towards urban centers
whereas illiterate people move to rural areas. Arif (2005) came up with
the same pattern of migrants using the PSES data.
Regarding the age of the migrants, Arif (2005) found that young
people mainly move towards urban areas whereas there is substantial
percentage of older people--around 30 to 35 percent--who move to rural
areas. These findings are in line with the earlier study by Irfan (1986)
who noted that young and educated people move towards urban areas and
illiterate people move towards rural areas.
Arif (2005) carried out occupational profile of migrants and
non-migrants by using the PSES data. The author found that majority of
the employed non-migrants was involved in agriculture, elementary
occupations and service sector. In-migrants also participated in
agriculture and elementary occupations along with crafts work.
Comparison of income profile revealed that in-migrants were slightly
better off than non-migrants. Further, rural households received almost
four times higher remittances as compared to their urban counterparts.
The effect of internal remittances, however, was noted to be
significantly marginal as compared to the remittances received by
international migration. The socio-economic status of the rural-rural
migrants remained unprivileged with higher percentage of stunted
children as compared to non-migrants' children [Arif (2005)].
Using the PIHS 1998 data, Memon (2005) found land ownership among
the important variable which reduces the probability of out-migration
from rural areas of the country. The author argued that land ownership
provide both economic and social capital, and hence increases both
monetary and social costs of migration from rural areas for land owners.
3. THEORIES ON MIGRATION
While reviewing theories on migration, Massey, et al. (1993)
divided theories on migration into three categories: micro theory
provides the description at individual level; macro theory draws the
picture at regional level; and miso theory presents the situation at
household level.
Neoclassical economics provide the description of both micro and
macro level theories. The micro level neoclassical migration theory
regards migration an outcome of rational decision of a person based on
cost-benefit analysis for positive return. Individuals include all sort
of physical, emotional, and psychological costs and benefits while
making their calculations. The macro level neoclassical economic theory
states that wage differentials--an outcome of disproportionate labor and
capital endowments--instigate migration between two regions. Due to
migration the wage differentials tend to minimize and migration seizes
up gradually as the wage differentials diminish. The new economics of
migration provide miso theory of migration and argues that it is not the
individual who takes decision based on personal cost and benefit rather
all relevant people--families or household members--decide collectively
about a migratory move.
Summing up, a number of studies have addressed the issue of
migration in the country, yet some issued remained unexplored. First,
all these studies used cross-sectional data which encounters some
methodological shortcomings for studying the determinants of migration.
The research on the determinants of migration in Pakistan lacks the use
of the panel data despite the availability of sufficient information on
this topic: two rounds of the PSES and three rounds of Pakistan Panel
Household Survey (PPHS). Second, all studies addressed issues pertaining
to the place of destination--inmigration--without any focusing on the
place of origin--out-migration. Both abovementioned data sets also
provide a module on out-migration.
4. RESEARCH OBJECTIVES
The overall objective of this study is to investigate the
determinants of out-migration in rural Pakistan. The study aims to
address the following research questions:
(1) Is there any influence of household poverty status on
out-migration?
(2) Do schooling, dependency ratio, household size, land ownership
and place of residence have any significant association with
out-migration?
(3) What is the role of household characteristics on external and
internal migration?
We embed our conceptual model of the study by combining
migration-inducing factors at all three levels proposed in micro, macro
and miso level theories: individuals decide to move while considering
all factors at personal, familial and regional levels.
5. DATA AND METHODOLOGY
The data for this study is used from the Pakistan Panel Household
Survey (PPHS). The survey covers 16 districts from all four provinces.
Thus far three rounds of this survey have been completed. These rounds
provide information on individuals and households residing in rural
areas of the selected districts. The last round in 2010, however, also
included urban areas in the sample. The PPHS provides detailed
information on all types of migrants: internal, external and returned.
This analysis uses information from two rounds of PPHS conducted in 2001
and 2010. However, as this survey targets only 16 districts, results
cannot be generalized for all the country.
PPHS defines a person as a migrant who leaves his place of
residence for a period of three months or more. The paper investigates
out-migration-both internal and external. External migrants are persons
who migrated ovearseas whereas those individuals who moved within
country are referred as internal migrants. The sample households are
located in rural areas only. The descriptive statistics reveal
out-migration from rural to both urban and rural areas. The external
migration is referred to out-migration of individuals to
abroad/overseas.
For this analysis, we applied a widely used standard logistic
regression model to look at association of household poverty status and
other household characteristics with out-migration, taking place between
2002 and 2010 in rural areas of Pakistan. We have investigated the
association of pre-migration household characteristics and
individual's age with out-migration. Out-migration has taken place
between 2002 and 2010, whereas the household characteristics- land
ownership, poverty status and others are collected from 2001 wave of
PPHS.
To investigate this association, we have used standadrd logistic
analysis with pooled data. The logistic regression is a nonlinear
regression (binary response) model specifically designed for binary
dependent variables. The logistic regression uses cumulative standard
logistic distribution. The coefficients of the logistic regression model
are estimated by maximum likelihood. The logistic fit maximum likelihood
models with dichotomous dependent variables coded as 0 and 1.
A general form of the model can be described as
Logit [P(y = 1)] = [[beta].sub.0] + [[beta].sub.1][X.sub.1] +
[[beta].sub.2][X.sub.2] +.... [[beta].sub.k][X.sub.k]
where y is a limited dependent/binary variable, [[beta].sub.0] is
constant, [X.sub.k] is vector of independent variables, and
[[beta].sub.k] represents a parameter estimate for the kth independent
variables. Dependent variable coded as 0 indicates the absence of the
characteristic, whereas coded as 1 indicates presence of the
characteristic. For instance, in this analysis, the dependent variable
out-migration takes on two unique values, 0 and 1. The value 0 denotes a
person ismot a migrant, and 1 denotes a person is a migrant.
The multinomial logistic regression is applied to investigate the
association of individual and household level characteristics with
out-migration. The multinomial logistic regression applies to maximum
likelihood models when dependent variable has more than two outcomes and
the outcomes do not represent any natural ordering. In this study, the
multinomial logistic regression is used to investigate the probability
of migrating to external (overseas) or internal location (within the
country) compared to no migration.
Let's assume that y is a dependent variable with three
outcomes 1, 2 and 3. A set of coefficients, [beta]s ([beta], (1) [beta],
(2) and [beta], (3)) are estimated relating to each outcome. Then the
mathematical form of the model for each outcome is as follows:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
By setting any coefficient, say [[beta].sup.(1)] = 0, the remaining
coefficients [[beta].sup.(2)] and [[beta].sup.(3)] will measure the
change relative to the y = 1 and so forth. The three coefficients
[[beta].sup.(1)], [[beta].sup.(2)] and [[beta].sup.(3)] will vary
because they have different interpretations.
6. RESULTS
6.1. Descriptive Statistics
Descriptive statistics on various characteristics of migrants and
non-migrants are reported in the following tables. As seen in Table 4,
overall only 2.6 percent of the household members (age 15+) migrated
during the period between 2002 and 2010 which include only 11 females.
The proportion of internal migrants is considerably more than the
external migrants (1.7 percent vs. 0.9 percent). As reported in Table 5,
majority of the migrants belong to younger age groups (under 30). The
proportion of internal and external migrants is highest among 21-25 and
26-30 years old respectively.
The decision to migrate is not necessarily an individual level. It
could be a household level decision as it has important implications for
household's resources, composition and economy. We have included
household level characteristics to examine their influence on
out-migration. These characteristics are average household size, average
years of schooling and average dependency ratio. As reported in the
Table 6, household size of external migrants' households is
considerably higher compared to households whose members migrated to
internal locations. It seems that larger households can afford to send
its members overseas compared to smaller households. In case of
households' stock of human capital (average years of schooling),
the evidence suggests that households with more (less) years of
schooling send their members abroad (internally).
The household dependency ratio could also encourage households to
send members out of home mainly for work. On average, there are more
dependents (children and elderly) than working age members in the
household as average dependency ratio of total sample is 105.0. The
households having more working age members (15-64) than dependents
(under 15 and 64+) are sending its members to overseas and internal
locations.
Household's economic (poverty) status is very important factor
in migration related decisions. On one hand, poor individuals/households
are expected to consider out-migration to explore better economic
opportunities and mitigate the effects of poverty. On the other hand,
poverty could also hinder the decision to out-migrate due to higher
costs associated with it. As seen in Table 7 below, prevalence of
out-migration is lower among poor (1.9 percent) compared to non-poor
(2.8 percent). Moreover, higher proportion of poor and non-poor decided
to move to internal locations compared to moving overseas. However, the
gap between moving overseas and internally is slightly wider in case of
poor migrants.
In the rural areas, land is one of the most valuable and most
important assets households aspire to hold. The results indicate that
lack of ownership of land is associated with out-migration (see Table
8). Those with no land have higher prevalence of out-migration
particularly to internal locations compared to those who own land.
Geography/location of household also plays a vital role in
out-migration. The residents in poor, deprived and under developed
regions/areas are expected to out-migrate. The proportion of external
and internal migrants is highest in Khyber-Pakhtunkhawa and Punjab
provinces respectively (see Table 9). Surprisingly, no one out-migrated
in Balochistan province, although the law and order situation in the
province has deteriorated over the period.
The comparison of out-migrants by land status of district of
origin, contrary to our expectations, shows that more individuals
out-migrated from agricultural districts compared to arid (3.1 percent
vs. 1.4 percent) and particularly to internal locations (see Table 10).
Amongst the districts, Bahawalpur has the highest proportion of
out-migrants followed by Vehari and Attock (see Table 11). Both
Bahawalpur and Vehari are agricultural districts unlike Attock. The
comparison of migrants by place of destination illustrates that the
proportion of external migrants is highest in Mardan followed by Dir and
Attock, whereas proportion of internal migrants is highest in Bahawalpur
followed by Vehari and Attock.
The analysis reveals that out-migrtion took place only for economic
reason. Thus, reason for migration was not included in the analysis.
Further, districts/regions like Swat and FATA which have observed
internally displaced persons (IDPs) were not part of PPHS, hence no
information is available on out-migration due to conflict.
6.2. Regression Results
We included various characteristics--socio-economic, demographic
and geographic location of the household--to capture their influence on
out-migration. These include household's average years of
schooling, household size, average dependency ratio of the household,
poverty status (2), land ownership, land status of the district of
origin, and place of residence before migration. (3) As some of the
individuals could not be matched across the two waves of the survey,
some of the individual-level variable, such as marital status,
education, employment status, were not included in the analysis. We
employed the logistic regression technique to find out predictive
variables of out-migration. The results indicate that average years of
schooling and household size have a significant positive association
with out-migration. Individuals living in more educated and large
households are significantly more likely to move out. Further, of
household's poverty and land ownership statuses- both reflect
economic position of the household, only land ownership turns out to be
a significant negative determinant of out-migration. A person from a
household with some owned land is significantly less likely to migrate
compared to a person from a household with no land ownership.
Further household characteristics related to geographic location
also appear significant determinants of out-migration. Individuals
residing in Sindh and Khyber-Pakhtunkhwa (KP) provinces are
significantly less likely, whereas those residing in agricultural
districts are significantly more likely to migrate compared to
individuals residing in Punjab province, and arid districts
respectively.
Further, we divided the move by destination: external (abroad) and
internal (within the country). We applied multinomial logistic
regression to investigate determinants of moving to external and
internal locations. Results indicate that compared to reference group
(no move), people are significantly more likely to move abroad and
internally with increase in their age, but for older people it is less
likely to move. Moreover, in comparison to those who did not move,
average years of schooling of the household members, household size, and
dependency ratio play a significant role in moving to overseas but not
in case of internal move. Increase in education level of the household
proved to be helpful in taking a decision for one of the
household's member to migrate abroad. Further, increase in
household size also improves the chance of migrating abroad by any
member of the household, whereas increase in dependency ratio in the
household significantly reduces the probability of moving abroad.
People with owned land, and living in Sindh have a significant
lower probability of moving abroad and within the country, whereas
living in KP has a significant higher (lower) probability of moving
abroad (within country) compared to reference groups. However, being a
resident of an agricultural district compared to arid district
significantly increases the probability of moving abroad.
7. CONCLUSION AND DISCUSSION
The objective of this research was to study the determinants of
out-migration in rural Pakistan using the panel data. We embed our study
in the new economic theories of migration which postulate that it is not
an individual's decision in isolation to migrate rather it is a
collective thought of family and household members. We therefore
included variable at household level in the analysis.
People are more likely to migrate with increase in the age but for
the older people it is less likely for them to move. This may imply that
for people it is more convenient to move when they are young. Further,
the analysis revealed that it is more likely for those households with
higher average year of schooling to have an out-migrant to overseas.
Similarly, greater household size also increases the probability of
migration abroad of any member of such households. Higher dependency
ratio, on the other hand, reduces the chances of movement of any member
of to overseas. Among the variables concerning the economic situation of
the households, land ownership reduces the chances of migrating, whereas
the poverty status of the household was found to have no role in
migration decisions.
Among the variables at macro level, the results of the analysis
revealed that those residing in other provinces are less likely to
migrate as compared to those residing in Punjab. This trend might be due
to the fact that Punjab is adjacent to all four provinces thereby
movement from Punjab to other provinces becomes quite economical. People
residing in the agricultural district were found to be more likely to
migrate as compared to those residing in the arid areas. People
migrating from the agriculture areas might be involved in agriculture
labor. These results are in line with what was revealed by PIHS 1998
data that land ownership reduces the probability of migration as
discussed by Memon (2005).
After reviewing recent research studies on internal migration in
the country we have observed some research gaps. First of all push
factors of internal migration have been ignored to some extent. Most of
the research has been conducted from the place of destination with a
very small focus on the previous or original place of migrants, thereby
ignoring those factors which made people move from the place of origin.
Further, in most of the studies the study population, internal
migrants in this case, has been studied from a
distance--quantitatively--without finding out how did they adapt to the
atmosphere in the place of destination; till now no qualitative study
has been designed to study the effects of migration on the overall life
conditions of the migrants. Further people who are left behind, the
impact of migration on these people is still yet to be studied.
Especially the effect of migration of male head of household on their
spouses and children, who find no one to take care of them, is worth
studying.
There are surveys which provide information on the place of origin,
thereby making it possible to study return migration in detail. However,
surveys like LFS which are conducted in succession should also include
such modules as has been done in PSES, PLM and PPHS. For example, the
inclusion of return migration, enables us to study those who return,
especially those were unable to adapt to the destination environment.
LFS provide information of migrants at their places of destination only
tell us success story of those who successfully adjusted at their place
of destination.
Various data sets came up with different results mainly due to
inconsistencies in the definitions. For example, definition of urban and
rural areas has been changed during various censuses, thereby causing
difficulties in studying pattern of rural-urban movements [Arif (2003)].
Distance of migration has been the focus of some studies.
Operationalisation of distance is, however, encountered by some
limitations. For example inter-administrative unit--be it a district or
province--is regarded as long distance movement without taking into
account the actual distance covered by the migrants. Global Information
System (GIS) along with other spatial econometrics techniques may be
employed to get a valid measure of distance of movement.
Pakistan has faced natural disasters, for example earthquake in
2005, floods and the internally displaced movement due to military
actions, quite frequently. In turn, the shape and dimensions of internal
movement in Pakistan might have disturbed a lot, and studying the
determinants using the conventional lens may overlook some important
features.
From the review it is somehow clear that migration has negative
impact on the well-being of the rural-rural migrant households in the
country, whereas there is hardly any evidence of economic benefit of
migration for other internally migrated households. These findings point
out that the decision to migrate was not based on a shrewd judgment, and
people who move might have taken the decision haphazardly. The decision
making process of migration that who decides to moves and how people
decide regarding their destination is worth investigating.
Migration issue has not been taken up by the policy makers in the
country. As the issue is closely related with poverty, Poverty Reduction
Strategy paper (PRSP)--an official document regarding poverty reduction
policy--does not address this issue substantially [Memon (2005)]. In the
country the policy regarding internal migration and urbanization is a
laissez-faire policy. The need of the hour requires a cogent proactive
policy which not only gives incentives but also imposes restrictions.
Policy makers should also intervene in the process of urbanization by
giving incentives for the people so that the main streams of migration
may be diverted from the main urban hubs. Establishing job opportunities
and educational institutions in small and medium size cities may serve
the purpose.
On the other hand there should be restriction on mobility of
people. A lucid way of achieving this objective is to have a stringent
registration system in contact. Everybody living in a locality should be
registered at an address. Further jobs may be given to those people who
belong to that particular area. As a matter of fact, provincial
government jobs do have a restriction of domicile. Finally illegal
squatter settlements should be discouraged. Above all, there is a need
for strong political will for enhancing well-being the people and urban
planning may simply be an offshoot of this policy framework.
REFERENCES
Arif, G. M. (2005) Internal Migration and Household Wellbeing: Myth
or Reality. Institute of Developing Economies. (IDE Discussion Papers
72).
Arif, G. M. (2003) Urbanization in Pakistan: Trends, Growth, and
Evaluation of the 1998 Census. In A. R. Kemal, M. Irfan and Naushin
Mahmood (eds.) Population of Pakistan: An Analysis of 1998 Population
and Housing Census. Islamabad. Pakistan Institute of Development
Economics.
Barkley, A.P. (1991) The Determinants of Inter-district Labour
In-migration in Pakistan, 1971-1980. Pakistan Institute of Development
Economics, Islamabad. The Pakistan Development Review 30:3, 275-296.
Chaudhry, M. J. (2003) Socio-economic and Demographic Profile of
Balochistan: An Analysis of the 1998 Population and Household Census.
Pakistan Institute of Development Economics. (Census Monograph Series
No.4).
Farah, N., M. I. Zafar and N. Nawaz (2012) Pakistan Journal of Life
and Social Sciences 10:1.
Farooq, M and M. A. Cheema (2005) Likelihood Impact of Internal
Migration on Family Institution in Punjab, Pakistan.
Farooq, M and Z. H. Javed (2009) The Impact of International
Migration on Migrants' Families Left Behind in the Rural Areas of
Pakistan. Pakistan Journal of Agriculture Science 46:4.
Farooq, M. A. Mateen and M. A. Cheema (2005) Determinants of
Migration in Punjab, Pakistan: A Case Study of Faisalabad Metropolitan.
Journal of Agriculture and Social Science 1:2.
Gazdar, H. (2003) A Review of Migration Issues in Pakistan. Paper
prepared for the Regional Conference on Migration, Development and
Pro-Poor Policy Choices in Asia, Dhaka, 22-24 June.
Hamid, S. (2010) Rural to Urban Migration in Pakistan: The Gender
Perspective. PIDE (Working Paper Series 2010:56).
Harris, J. R. and M. P. Todaro (1970) Migration, Unemployment and
Development: A Two-sector Analysis. American Economic Review 60,
126-142.
Irfan, M. (1981) Pakistan's Labour Force, Its Size and
Structure: A Review of Evidence. Pakistan Institute of Development
Economics (Research Report No. 116).
Irfan, M., L. Demery, and G. M. Arif (1983) Migration Patterns in
Pakistan: Preliminary Results from the PLM Survey, 1979. Studies in
Population, Labour Force and Migration Project. Pakistan Institute of
Development Economics, Islamabad. (Report No. 6).
Karim, M. S. and Abu Nasar (2004) Migration Patterns and
Differentials in Pakistan: Based on the Analysis of 1998 Census Data. In
A.R. Kemal, M. Irfan and Naushin Mahmood (eds.) Population of Pakistan:
An Analysis of 1998 Population and Housing Census. Islamabad. Pakistan
Institute of Development Economics.
Khan, A. H. and L. Shenaz (2000) Determination of Internal
Migration : Evidence from the LFS 1996-7. The Pakistan Development
Review 39:4, 695-712.
Khattak, N. (2003) Socio-economic and Demographic Profile of NWFP:
An Analysis of the 1998 Population and Household Census. Pakistan
Institute of Development Economics. (Census Monograph Series No.2).
Mahmud, M., T. Musaddiq and F. Said (2011) Determinants of Internal
Migration in Pakistan -Lessons from Existing Patterns. Presented at 26th
AGM and Conference of Pakistan Society of Development Economists, held
at Islamabad.
Mehta, G. S. (1991) Socio Economic Aspects of Migration. Deep and
Deep Publications: New Delhi, India.
Memon, R. (2005) Determinants and Trends of Internal Migration in
Pakistan. Institue of Developing Economies. (IDE Discussion Papers 72).
Naeem, J. (2003) Socio-economic and Demographic Profile of Punjab:
An Analysis of the 1998 Population and Household Census. Pakistan
Institute of Development Economics. (Census Monograph Series No.l).
Oda, H. (2007) Dynamics of Internal and International Migration in
Rural Pakistan. Asian Population Studies 3:2, 169-179.
Rukanuddin, A. R. and M. A. Chaudhry (2003) Socio-economic and
Demographic Profile of Sindh: An Analysis of the 1998 Population and
Household Census. Pakistan Institute of Development Economics. (Census
Monograph Series No.3).
Comments
The study provides a panel analysis of the out migration by using
the Pakistan Panel Household Survey (PPHS). Also, the study proclaims
that since the decision to migrate is done by a family; therefore it
incorporates the household's level characteristics in the analysis.
In particular, the focus is on the role of the ownership of land and the
level of poverty in the decisions to migration. The paper studies an
important issue in the sense that migration has become an important
phenomenon in the Pakistani society, in particular after the recent
surge of the Internally Displaced Persons (IDPs). However, one should be
very careful in generalising the results of the study as it still lacks
on several technical grounds.
The paper lacks consistency and congruence. For instance, the aim
of the paper is to investigate the out migration; but at the same time,
it introduces the concepts like internal and external migration without
differentiating between these concepts. Besides, the paper reports three
objectives; however, neither of them is sufficiently discussed or
conformed in the results. The study also claims that the migration might
affect the individual's or household's wellbeing; however,
again, there is no indication of this claim in the results section.
Also, the paper claims that it uses two rounds of PPHS (2002 and 2010);
but one does not see any panel regression elsewhere in the paper. It
would really make things easy if the authors could explain about the
reported method, i.e. it should be clear whether the regressions are
fixed effect logistics, random effect logistics or just pooled? To me,
it is just simple logistic analysis with pooled data; however the
authors should explicitly explain it. Additionally, the PPHS only
reports migration data for a small fraction of the households; so it
should be explained how they handled the problem of missing data.
As stated earlier, the purpose of the study is to analyse the
impact of poverty, land values, and the other social and economic
characteristics on out-migration in rural Pakistan. However, the study
fails to highlight all of the important controls in this regard. For
instance, a detailed discussion of the chosen and the possible omitted
controls, in particular in the light of existing literature would be
highly appreciated. In particular, the authors must elaborate on the
level of the study, i.e. whether it is individual level or
households' level? For instance, I see age as one of the variables
in the regressions, so it seems to me to be the individual level.
However, the authors should explicitly highlight this fact. If the study
is individual level, then some of the households' level
characteristics such as poverty status would be repeated for members of
the each household. In such a case the cluster standard errors are
appropriated, i.e. the authors should report it. In the study, the
authors used logistic regression while, at the same time, it also
introduces the multinomial logistic regression. They should eliminate
such redundancy from the study.
Rural-urban Migration is the function of economic incentives. Most
of the migration is done in the context from rural to urban areas and
not from the perspective of rural to rural areas. Major reasons are
economic, marriage and family. However, in the results, the status of
poverty is not significant which is a very strong statement and contrary
to the incentives theory of migration. At the same time, the ownership
of land is significant. The author should look at the fact that it might
be the case that the ownership of land is highly correlated with the
status of poverty. Similarly, the dependency ratio is insignificant
while the households' size is significant. Again, it may be the
case that the households' size and the dependency ratio are highly
con-elated almost in every society.
The ownership of land may be endogenous as higher remittances might
affect the purchase or values of land and remittances are normally the
function of migration. Besides, there is no control for civil conflicts
in the study. However, it is very important given the recent impact of
IDPs on migration. Similarly, the decision to migrate and other
important decisions such as the schooling decisions or job-market
decisions might be simultaneously determined. This type of simultaneity
bias should be taken care of. The paper should be revised in the light
of these comments.
Karim Khan
Pakistan Institute of Development Economics, Islamabad.
Rizwan ul Haq <r.ul.haq@pide.org.pk> is Research Demographer,
Ajmal Jahangeer <ajmal@pide.org.pk> is Staff Economist and Azkar
Ahmad <azkar@pide.org.pk> is Research Economist at the Pakistan
Institute of Development Economics, Islamabad. The views expressed in
this paper are solely those of the authors.
(2) As calculated by Arif and Shujat (2012).
(3) Average years of schooling, household size, dependency ratio,
poverty status, land ownership and residence related characteristics are
of 2001.
Table 1
Percentage Distribution of Internal Migrants by Direction of Move
2000-01
PSES (all 1996-97 LFS 1998-99 P1HS 1979 PLM *
Direction of ages-both (Age 10 + - (Male- Punjab (all ages--
Move sexes) both sexes both sexes) both sexes)
Urban-urban 19.5 43.0 22.7 14.9
Urban-rural 5.9 6.9 6.8 13.9
Rural-urban 38.8 29.8 40.7 29.8
Rural-rural 36.2 20.3 29.7 41,3
All 100 100 100 100
Source: Arif (2005).
Table 2
Number and Percent of Life Time Migrants in Pakistan and
Their Place of Origin, 1951-1998
Year
1951 1961 1973
Total Life-time Migrants 7,755,402 8,777,746 10,129,993
Total Internal Migrants 1,397,285 2,826,036 4,436,316
(100) (100) (100)
Within Province 953,074 1,937,052 2,578,734
(68.2) (68.5) (58.1)
Other Provinces 444,211 888,984 1,857,582
(31.8) (31.5) (41.9)
Year
1981 1998
Total Life-time Migrants 9,959,251 10,829,264
Total Internal Migrants 5,172,576 8,368,723
(100) (100)
Within Province 3,436,086 5,705,447
(66.4) (68.2)
Other Provinces 1,736,490 2,663,276
(33.6) (31.8)
Source: Karim and Nasar (2004); Note: Figures in parentheses are
percentages.
Table 3
Reported Reasons for Migration
Percentage of Migrants
LFS PIHS LFS Census
Reason for Migration 1996-97 1998 1997-98 1998
Job transfer 7.1 2.8 5.5 12.1
Finding a job 10.0 12.5 8.9 NA
Business 4.3 2.8 4.4 8.8
Education 0.9 1.1 0.5 1.2
Health 0.2 0.3 0.2 0.01
Marriage 26.1 41.2 26.1 17.0
With family 19.8 22.3 23.7 42.8
Return home 6.7 3.1 9.3 1.1
Independence NA 8.4 NA NA
Others 25.0 5.7 21.4 16.9
Proportion of economic migrants
in migrants sub-sample 21.3 18.1 18.8 20.9
Proportion of economic migrants
in full sample 2.3 4.7 1.7 1.7
Source: Khan and Shahnaz (2000); Memon (2005).
Table 4
Distribution of Out-Migrants by Gender
External Internal
Non-migrants Migrants Migrants Total Sample
No. % No. % No. % No. %
Female 4572 99.8 1 0.0 10 0.2 4583 100.0
Male 4688 95.3 85 1.7 147 3.0 4920 100.0
Both Sexes 9260 97.4 86 0.9 157 1.7 9503 100.0
Table 5
Distribution of Out-migrants and Non-migrants by Age at Time of
Out-migration
External Internal
Non-migrants Migrants Migrants Total Sample
No. % No. % No. % No. %
<20 4476 97.9 13 0.3 85 1.9 4574 100.0
21-25 893 93.9 23 2.4 35 3.7 951 100.0
26-30 692 94.9 24 3.3 13 1.8 729 100.0
31-40 1170 97.9 20 1.7 5 0.4 1195 100.0
41-50 894 98.7 4 0.4 8 0.9 906 100.0
51-60 598 99.5 2 0.3 1 0.2 601 100.0
61 & above 537 99.8 0 0.0 1 0.2 538 100.0
All ages 9260 97.5 86 0.9 148 1.6 9494 100.0
* Non-migrants' age is as of 2001, whereas age of migrants is at the
time of first migration.
Table 6
Average Household Size, Years of Schooling and Dependency
Ratio before Out-migration (in 2001)
External Internal
Non-migrants Migrants Migrants Total Sample
No. Avg. No. Avg. No. Avg. No. Avg.
Average 9260 10.5 86 12.6 157 8.9 9503 10.5
household
size
Average years 9260 2.2 86 3.3 157 2.3 9503 2.2
of schooling
Average 9241 105.7 86 79.7 157 98.1 9484 105.0
dependency
ratio
Table 7
Distribution of Out-migrants by Poverty Status in 2001
External Internal
Non-migrants Migrants Migrants Total Sample
No. % No. % No. % No. %
Non-poor 7072 97.2 79 1.1 124 1.7 7275 100.0
Poor 2308 98.0 12 0.5 34 1.4 2354 100.0
All 9380 97.4 91 0.9 158 1.6 9629 100.0
Table 8
Distribution of Out-migrants by Land Ownership Status before
Out-migration in 2001
External Internal
Non-migrants Migrants Migrants Total Sample
No. % No. % No. % No. %
No land 3715 95.5 65 1.7 109 2.8 3889 100.0
Own land 5433 98.8 20 0.4 45 0.8 5498 100.0
All 9148 97.5 85 0.9 154 1.6 9387 100.0
Table 9
Distribution of Out-migrants by Province of Origin
External Internal
Non-migrants Migrants Migrants Total Sample
No. % No. % No. % No. %
Punjab 3159 95.3 30 0.9 127 3.8 3316 100.0
Sindh 2787 98.8 9 0.3 26 0.9 2822 100.0
Khyber- 1975 97.4 51 2.5 1 0.0 2027 100.0
Pakhtunkhawa
Balochistan 1330 100.0 0 0.0 0 0.0 1330 100.0
All 9251 97.4 90 0.9 154 1.6 9495 100.0
Table 10
Distribution of Out-migrants by Land Status of District of Origin
External Internal
Non-migrants Migrants Migrants Total Sample
No. % No. % No. % No. %
Arid 3103 98.5 32 1.0 14 0.4 3149 100.0
Agricultural 6173 96.9 58 0.9 140 2.2 6371 100.0
All 9276 97.4 90 0.9 154 1.6 9520 100.0
Table 11
Distribution of Out-migrants by District of Origin
External Internal
Non-migrants Migrants Migrants Total Sample
No. % No. % No. % No. %
Faisalabad 459 96.6 7 1.5 9 1.9 475 100.0
Attock 413 95.2 7 1.6 14 3.2 434 100.0
Badin 1030 99.1 1 0.1 8 0.8 1039 100.0
Dir 1108 97.8 25 2.2 0 0.0 1133 100.0
Hafizabad 635 96.2 5 0.8 20 3.0 660 100.0
Vehari 659 95.1 10 1.4 24 3.5 693 100.0
Muzaffar Garh 388 97.5 0 0.0 10 2.5 398 100.0
Bahawalpur 619 92.4 1 0.1 50 7.5 670 100.0
Nawab Shah 630 98.9 3 0.5 4 0.6 637 100.0
Mirpur Khas 409 97.8 2 0.5 7 1.7 418 100.0
Larkana 727 98.6 3 0.4 7 0.9 737 100.0
Mardan 617 95.8 26 4.0 1 0.2 644 100.0
Lakki Marwat 251 100.0 0 0.0 0 0.0 251 100.0
Loral ai 46.8 100.0 0 0.0 0 0.0 468 100.0
Khuzdar 399 100.0 0 0.0 0 0.0 399 100.0
Gwadar 464 100.0 0 0.0 0 0.0 464 100.0
All 9276 97.4 90 0.9 154 1.6 9520 100.0
Table 12
Logistics Regression Estimates of Out-migration in Rural Pakistan
Out-migration Coef. Std. Err.
Age 0.284 * 0.032
Age square -0.005 * 0.001
Average years of schooling 0.066 ** 0.031
Household size 0.025 * 0.01
Dependency ratio -0.001 0.001
Poverty status -0.199 0.174
Household owns land -1.376 * 0.148
Sindh -1.550 * 0.192
Khyber-Pukhtunkhwa (KPK) -0.842 * 0.204
District status 0.645 * 0.21
Constant -6.238 * 0.479
N = 9391; * significant at 0.01, ** significant at 0.05 and ***
significant at 0.10. Reference categories are: non-poor, no land
ownership, residence in Punjab, and in arid districts.
Table 13
Multinomial Logistic Regression Estimates of Out-migration in Rural
Pakistan
Move Abroad Move Internally
Characteristics Coef. Std.Eror Coef. Std.Eror
Age 0.465 * 0.067 0.249 * 0.043
Age square -0.007 * 0.001 -0.005 * 0.001
Average years of 0.153 * 0.047 -0.012 0.045
schooling
Household size 0.037 * 0.011 0.007 0.02
Dependency ratio -0.004 ** 0.002 0.000 0.001
Poverty status -0.377 0.328 -0.110 0.217
Household owns land -1.804 * 0.258 -1.263 * 0.194
Sindh -1.361 * 0.409 -1.793 * 0.243
Khyber- Pukhtunkhwa (KP) 0.849 * 0.273 -4.763 * 1.027
District status 0.753 * 0.268 0.110 0.32
Constant -11.083 * 1.052 -4.980 * 0.645
N = 9378; reference category is no move; * significant at 0.01, **
significant at 0.05 and *** significant at 0.10. Reference categories
are: non-poor, no land ownership, residence in Punjab, and in arid
districts.
