How has internal migration in Albania affected the receipt of transfers from family and friends? Florian Tomini Maastricht Graduate School of Governance Maastricht University Jessica Hagen-Zanker Overseas Development Institute, UK World Bank Conference on Poverty and Social Inclusion in the Western Balkans Brussels, Belgium, December 14-15, 2010
24
Embed
How has internal migration in Albania affected the receipt of transfers from family and friends? Florian Tomini Maastricht Graduate School of Governance.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
How has internal migration in Albania
affected the receipt of transfers from family
and friends?
Florian Tomini Maastricht Graduate School of Governance
Maastricht University
Jessica Hagen-ZankerOverseas Development Institute, UK
World Bank Conference on Poverty and Social Inclusion in the Western BalkansBrussels, Belgium, December 14-15, 2010
Maastricht Graduate School of Governance
Motivation
• Family networks economic, social support etc
• Networks non-stable affected by social, economic changes, or physical location
• What happens to the network when migration is characterized by relocation of households?
– Will the composition of received transfers change?
– Will the sending relatives be different?
• Our study: looks at how family solidarity & networks have been affected by internal migration when entire households move.
• Contacts and support after migration (Litwak 1960; Jitodai 1963; Wellman et al 1997; Ruan et al 1997).
• Solidarity after transition (Cox 1996; Vullnetari & King 2008).
• Other views co-insurance agreements (Stark 1991).
Maastricht Graduate School of Governance
Internal migration and Albania
• Between 1945-1990 internal migration in Albania was centrally controlled (international migration not allowed).
• The collapse of communist regime in 1990 people migrated either internationally or internally.
• Internal migration is not circular and often characterized by relocation of the whole household (De Soto et al, 2002; Cila, 2006) .
• Motivation seem to relate mostly to economic reasons (work seeking, etc) (Carletto et al., 2004).
Maastricht Graduate School of Governance
Internal migration and Albania
Source: Based on Albania LSMS Data 2002 - 2005
025
000
5000
0#
of m
igra
nts
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
Year moved to current residence
Internal International
Maastricht Graduate School of Governance
Survey
• Hh-survey in peri-urban Tirana, April 2008
• Recently populated areas with high informality
• 112 hhs sampled, 26 also qualitative interview
• Sampling methodology:
– 1) Define the recently populated areas (5 main zones).
– 2) Sub-divide SU of 1 km2 within these zones using satellite maps.
– 3) Randomly select hhs within selected sub-sections
Maastricht Graduate School of Governance
Survey
• Migrant households come from nearly all districts, but especially from the Northern and Central mountainous areas (the darker areas on the map).
Maastricht Graduate School of Governance
Data – selection of family members
• Members of kinship (including relatives/non-relatives) they have been in contact with both now and in the past.
– Total: 1064 kinship members hhs are in contact with.
• Next getting the information on transfers with randomly selected (alphabetical order of given names) relatives/non-relatives:
- Parents/ parents in law (1)
- Siblings (2)
- Children (2)
- Other relatives (2)
- Non-relatives (friends, neighbours, etc) (1)
Maastricht Graduate School of Governance
Data - transfer questions
• Transfers to the household in the past 12 months.
• Hh are also asked about transfers in the past. – Transfers in 1991 if they moved before & in 1997
– Transfers in 1997 if they moved in 1998 or after.
• “Transfers” included:
– Financial transfers
– Transfers of goods
– Services transfers
Maastricht Graduate School of Governance
Methodology
• The transfers occur within a defined limit of time, and probabilities of consecutive transfers are not dependent on each other.
• Frequency data shows over-dispersion (variance is greater than mean) standard Poisson model not suitable
• Two may be the causes of this over-dispersion:– 1) idiosyncratic and random bias in receiving transfers
(households do not have the same probability for receiving a certain frequency of transfers), and
– 2) households do not receiving transfers systematically because of their characteristics or relatives characteristics (i.e. limited contacts in the past 12 months before migration).
Maastricht Graduate School of Governance
Methodology – model testing
• Models considered:– PRM (Poisson)
– ZIP (Zero Inflated Poisson)
– NBRM (Negative Binomial Regression)
– ZINB (Zero Inflated Negative Binomial)
• Results confirm over-dispersion due to idiosyncratic factors and random bias.
– NBRM and ZINB give best results.
-.1-
.05
0.0
5.1
Obs
erve
d-P
red
icte
d
0 1 2 3 4 5 6 7 8 9Count
PRM NBRM
ZIP ZINB
Note: positive deviations show underpredictions.
All transfers combined
Maastricht Graduate School of Governance
Methodology
• To account for over-dispersion among the count outcomes we use a “negative binomial regression model”, where:
iy
iii
ii y
yyY
!)(
)()Pr( ny ,...3,2,1
is the estimated value of the model dependent on a vectors of covariates,
- accounts for the over-dispersion in the data.
)exp()( iii xyE
- expected value of the model. - vector of estimated coefficients,
- vectors of covariates including characteristics of receiving household and sending relative.
i
ix
iy
1
Maastricht Graduate School of Governance
Empirical strategy
• We pool the data from before and after migration, accounting for when the transfer takes place with the “migration dummy”.
– When applicable, the variable is adjusted to the period before migration (i.e. age, number of children etc.).
• Models are estimated separately for different types of transfers and for all transfers combined.
• In addition, to check for how role of relatives has changed before and after migration we check for differences in coefficients using “seemingly unrelated estimations” (Weesie, 2000).
Maastricht Graduate School of Governance
Results – NBRM
Financial transfers Good transfers Service transfers Coef. st. error Coef. st. error Coef. st. error
Transfer after migration 1.24*** 0.32 -0.89*** 0.28 -0.93*** 0.28
Relative parent -0.07 0.62 1.04* 0.56 -1.14* 0.60
Relative child -0.57 0.86 1.76*** 0.64 0.57 0.69
Relative sibling 0.30 0.42 0.71* 0.39 -0.78* 0.43
Relative other -0.13 0.48 0.05 0.40 -1.78*** 0.45
Age hhh -0.03** 0.01 -0.02 0.01 -0.01 0.01
Gender hh head 1.66** 0.66 -0.68 0.66 -0.14 0.83
Education years hhh -0.05 0.06 0.09*** 0.03 0.09* 0.05
Hh extended family 0.36 0.29 -0.60** 0.29 -0.63** 0.28
Number of children hh -0.19 0.15 -0.03 0.12 0.37*** 0.14
Hh moved before 1997 -0.85*** 0.30 0.18 0.27 -0.18 0.27
Age relative/ friend 0.02* 0.01 -0.00 0.01 -0.01 0.01
Ln alpha 2.23*** 0.13 2.27*** 0.08 2.41*** 0.08 Number of observations 940 927 907 Pseudo R2 0.07 0.03 0.02 LR Chibar2 1353.83 6436.38 16000.00 P-value Chibar2 0.00 0.00 0.00
* significant at 10%; ** significant at 5%; *** significant at 1%
Maastricht Graduate School of Governance
Results - Predicted transfers before and after migration
0.5
11
.5P
red.
freq
. FIN
AN
CIA
L tr
ans
f.
20 40 60 80Age of hhh
Transfer after migration Transfer before migration
12
34
5P
red.
freq
. GO
OD
tran
sfe
r
20 40 60 80Age of hhh
Transfer after migration Transfer before migration
2.1 Financial transfers 2.2 Good transfers
24
68
10
Pre
d. fr
eq. S
ER
VIC
E tr
ans
f.
20 40 60 80Age of hhh
Transfer after migration Transfer before migration
51
01
5P
red.
freq
. ALL
tran
sf.
20 40 60 80Age of hhh
Transfer after migration Transfer before migration
2.3 Service transfers 2.4 All transfers
Maastricht Graduate School of Governance
Results - NBRM
• At all ages financial transfers are more frequent after migration, while services and goods are less frequent.
• Friends transfer more frequently financial transfers and services than other kinship members (effect not significant), but less goods. (Migration effect is not yet known).
• Frequency of financial transfers is higher from old to young hhh, and from male to female headed hh.
• Education of hhh influences negatively financial transfers, but positively other transfers.
• (Income variable influences transfers negatively but it is not significant.)
Maastricht Graduate School of Governance
Results – Migration effect on network
Differences in coeff. from separate NBRM (before & after migration)
Financial transfers Good transfers Service transfers
• The frequency of financial transfers from siblings and other relatives decreases if compared to the frequency of transfers from friends (same effect for parents but not significant).
• Similar trends are confirmed for services. Friends start transferring more frequently than parents, siblings and other relatives.
• Transfers from children increase more than form friends for financial transfers and goods (results are to be treated with caution).
Maastricht Graduate School of Governance
Conclusion
• Internal migration seems to have a positive effect on the receipt of financial transfers.
• Migrants receive less frequently goods (the change in types of goods exchanged), and services (more time spent in employment or job-search activities).
• Internal migration has affected the support network (transfers from friends and children have increased more than transfers from siblings and others).
• Caveats:
– Small-scale household survey in a very specific context
– Survey focused exclusively on migrant households
Maastricht Graduate School of Governance
Thank You!
Maastricht Graduate School of Governance
Transfer frequency from different types of kin
Type of kin the hh receives transfers from
Parents &
parents in law
Children Siblings Relatives Friends Total
Frequency financial transfer before migration
0.29 0 0.25 0.66 0.04 0.34
Frequency financial transfer in past 12 months
0.5 0.17 0.68 0.42 0.92 0.6
Frequency goods transfer before migration
3.26 0.7 3.5 2.18 2.36 2.89
Frequency goods transfer in past 12 months
3.16 2.56 2.39 1.62 1.26 2.18
Frequency services transfer from before migration
11.26 14.38 10.88* 4.79*** 7.93 9.11
Frequency services transfer in past 12 months
8.81* 12.89*** 7.08 3.35*** 6.73 6.65
Number of observations 61-151 18-54 182-407 110-235 25-132 397-987 Stars indicate whether the mean for each group is significantly different from the total mean
(* significant at 10%; ** significant at 5%; *** significant at 1%)
Maastricht Graduate School of Governance
Characteristics of migrant hh
• 75% nuclear families
• Average hh size >5
• >50% of hhhs completed primary & secondary school
Employment level for adult hh members
38%
7%20%
10%
10%
15%
Employed full- time
Employed part- time
Unemployed
Housewife
Student
Retired
Maastricht Graduate School of Governance
Hh family members/ close friends hhhs are in contact with
Relationship to hhh
10%
6%
39%
26%
19% Parents & parents inlaw
Children
Siblings
Other relatives
Friends/ neighbours
274274
417417
1101106161202202
Total: 1064 family & friends hh is in contact with