1
Migration and Remittances in Emerging Market Economies of
Southeast Asia: Do they Offer Paths for Structural Poverty
Transitions?
Arnoldshain Seminar XI“Migration, Development, & Demographic Change –
Problems, Consequences, Solutions”
Mulubrhan Amare
(with Herman Waibel & Lena Hohfeld )
Institute of Development and Agricultural Economics, School of Economics &
Management, Leibniz Universität Hannover, Germany
Outline
Background Objectives Conceptual framework Data description Econometric approach Results Summary & conclusions
2
Motivation (I)
The share of remittances to total income growing in many developing
countries.
More than 30-40% rural income from remittances (Nguyen et al., 2007)
Evidences on poverty reduction and inequality impact is mixed
Migrant in low quality employment
Remittances may spend for consumption
Increase inequality in the community of origin
Background
3Background Objectives Conceptual framework Data Econometric approach Results Conclusions
Motivation (II) Limitations on the impact of migration and remittances
1. Do not distinguish whether remittances lead to structural transition due
to asset growth or
2. Stochastic transition due to apparent higher income
3. Overestimation of poverty transition impact of migration & remittances
To learn the impact of migration on structural transitions, asset based
approach is used
Suitable for identifying structural impacts of migration and forward-
looking policy
Background
4Background Objectives Conceptual framework Data Econometric approach Results Conclusions
1. To investigate to what extent migration & remittances improve asset
accumulation in the source communities
2. To examine whether migration & remittances has impact on structural
poverty transitions of rural households
3. To identify channels though which migration & remittances can
potentially influence welfare dynamics of rural households
Objectives
5Background Objectives Conceptual framework Data Econometric approach Results Conclusions
Conceptual framework
6Background Objectives Conceptual framework Data Econometric approach Results Conclusions
Income is the product of households’ asset endowments & the returns to
capital (Barrett,2005) (1) itititritKitY it ]['
Y : income of household i in time t itK : a vector of household capital.
r : expected returns : an exogenous shocks
: transitory unearned income : the measurement error
Period specific income (structural income)
(2) )('][ itritKitYE
Conceptual framework
7Background Objectives Conceptual framework Data Econometric approach Results Conclusions
Total differentiation of the income equation (1)
(3) itddditdritKitritdKitdY ititit )][(']['
Taking the expectation of equation (3)
(4) itdritKitritdKitdYE ''][ : the structural income growth
Channels
1. Remittances: increase factor productivity
(5) ),,( vtGitKIfitr
Conceptual framework
8Background Objectives Conceptual framework Data Econometric approach Results Conclusions
2. Remittances: strengthen endowment effect
Asset growth can be expressed as a reduced form function
(6) ),,,,,,.( ,,
ititvtvtititititit ddGGdIIdKKfdY
Hypotheses
Remittances lead to greater asset growth for initially wealthy households,
,0/ itit dIdY
Remittances can result in a lower asset growth for poor households, 0/ itit dIdY .
Remittances impact welfare dynamics though its effect on factor productivity
Panel survey to assess vulnerability to poverty in
Asia
3 provinces in Vietnam (220 villages & 2200
households )
3 provinces in Thailand (220 villages & 2200
households )
2007, 2008 & 2010 of household & village level
surveys.
Across all three rounds, 2108 & 2095 households
appear in all rounds in Vietnam & Thailand,
Data description
9Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
1. Constructing asset indexitiptvtgkj iktijtjkj ijtjit GAAAL ,1 ))(()()6(
where livelihood L defined as household i income per capita divided by the nationally
defined poverty line tP in period t
The Econometric approach
10Background Objectives Conceptual framework Data Econometric approach Results Conclusions
2. Asset growth equation (7a) ivtititititit XZAMA 154131211 .
(7b) ivtititititit XZAIA 154131211
where itA refers to growth in asset; 1itM refer to migration decision; 1itI log of remittances
1itZ are a vector of household characteristics : it referees income and asset severity of shocks
1vtX are vectors of geographic location and natural variables
2. Asset growth equation
The Econometric approach
11Background Objectives Conceptual framework Data Econometric approach Results Conclusions
1. Compare asset growth between non-migrant and migrant households in the
community of origin as specified in equ. (7a).
2. Investigate the impact of remittances in explaining rural households’ asset
accumulation path equ. (7b).
3. Estimate both equ. (7a) & (7b) to test whether migration & remittances
impact the structurally poor households differently than non-poor ones.
The Econometric approach
12Background Objectives Conceptual framework Data Econometric approach Results Conclusions
3. Decomposition Analysis
Blinder–Oaxaca decomposition is employed to identify channels though
which migration & remittances can affect welfare dynamics
U
N
o
M
o
C
N
j
M
j
N
j
E
N
j
M
j
M
j
NM XXXGG )()()()8(
E : growth differences due to migrant households having lower endowments
C : growth differences due to migrant households having lower productivity
(marginal returns).
U : the growth differences unexplained by endowments or productivity
Results & discussions
13Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
For Vietnam
Year 2007 2008 2010 Average
Migrant households (%) 39 37 39 38
Number of migrant 3.28(2.29) 2.99(2.08) 3.1(2.15) 3.13(2.18)
Annual income from remit. 312(562) 481(629) 567(963) 453(732)
For Thailand
Migrant households (%) 45 44 47 45
Number of migrant (%) 4.31(2.56)
4.35(2.65) 4.33(2.79) 4.34(2.67)
Annual income from remit. 974(460) 1006(500) 1268(762) 1086(583)
Migrant households, income from remittance & number of migrants
Results & discussions
14Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
Descriptive statistics of households by migration status in Vietnam &Thailand
Vietnam Thailand
Migrant
(38)
Non-
migrant (62)
Difference
Test
Migrant
(45)
Non-migrant
(55)
Difference
Test
Monthly structural inc. 127(80) 115(84) *** 165(375 172(133)
Structural poor (%) 14 19 *** 18 15 **
HHsize 4.46(1.76) 4.33(1.76) 4.12(1.77) 3.97(1.71) ***
Children 0.19(0.19) 0.28(0.22) *** 0.19(0.16) 0.21(0.18) ***
Elderly 0.07(0.17) 0.12(0.25) *** 0.07(0.13) 0.11(0.22) ***
Mean edu. 8.81(3.04) 7.27(3.51) *** 6.45(2.42) 5.90(2.45) ***
Gender (%) 78 76 * 72 73
Ethnic (%) 86 75 *** - -
Membership(%) 91 85 *** 15 15
Results & discussions
15Background Objectives Conceptual framework Data Econometric Approach Results Conclusions Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
The Impact of Migration on Asset Growth by Welfare Status Vietnam
Variables All sample Structural poor Structural non-poor
Migrant household 0.020(0.015) 0.048(0.060) 0.007(0.014)
Household characteristics Yes Yes Yes
Village characteristics Yes Yes Yes
Severity of shocks Yes Yes Yes
Thailand
Variables All sample Structural poor Structural non-poor
Migrant household -0.042**(0.018) -0.080**(0.037) -0.028(0.018)
Household characteristics Yes Yes Yes
Village characteristics Yes Yes Yes
Severity of shocks Yes Yes Yes
The Impact of Remittances on Asset Growth by Welfare Status in Vietnam
Results & discussions
16Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
Variables All sample Structural poor Structural non-poor
Remittances 0.024***(0.006) 0.081***(0.031) 0.007(0.007)
Initial assets & household characteristics Initial asset index -0.502***(0.022) -0.761***(0.051) -0.505***(0.024)HH size -0.152***(0.020) -0.557***(0.114) -0.117***(0.017)Gender 0.038**(0.019) 0.029(0.076) 0.037**(0.016)Ethnic 0.177***(0.023) 0.146**(0.074) 0.168***(0.023)Membership 0.077***(0.020) 0.119**(0.057) 0.064***(0.022)Village level geographic location &natural conditions Mountainous -0.018**(0.005) -0.052***(0.027) -0.035**(0.014)Paved road 0.104***(0.014) 0.095*(0.056) 0.113***(0.014)Violence -0.058**(0.023) -0.120(0.075) -0.039(0.024)Epidemics -0.062**(0.024) -0.079(0.068) -0.065**(0.026)Public water supply 0.064***(0.015) 0.012(0.063) 0.080***(0.015)HHs electricity 0.000(0.000) 0.001(0.001) -0.000(0.001)No. of enterprises 0.001(0.003) 0.219(0.439) 0.001(0.003)HHs sanitation -0.000(0.000) 0.004**(0.002) -0.001***(0.000)Time to the market -0.001**(0.000) -0.001(0.001) -0.001***(0.000)Severity of shocks Asset loss -0.077***(0.026) -0.193**(0.077) -0.049***(0.017)Income loss -0.000(0.000) -0.001(0.002) -0.000(0.000)
Results & discussions
17Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
Variables All sample Structural poor Structural non-poor
Remittances -0.013**(0.005) -0.030**(0.014) -0.009(0.006)
Initial assets &household characteristics Initial asset index -0.562***(0.019) -0.920***(0.036) -0.670***(0.023)HH size -0.179***(0.021) -0.421***(0.060) -0.132***(0.028)Gender -0.170***(0.016) -0.403***(0.071) -0.201***(0.021)Membership 0.080***(0.022) 0.140**(0.064) 0.056**(0.024)Village level geographic location &natural conditions
Paved road 0.039*(0.020) 0.116**(0.048) 0.026(0.031)Violence -0.065***(0.020) -0.060(0.047) -0.093***(0.036)Epidemics -0.017**(0.012) -0.027(0.050) -0.040(0.037)Public water supply 0.018**(0.011) -0.056(0.050) 0.043(0.034)HHs electricity 0.000(0.002) -0.010*(0.006) 0.003(0.004)No. of enterprises 0.018(0.011) 0.031(0.028) 0.001(0.019)HHs sanitation 0.001***(0.000) 0.000(0.000) 0.001***(0.000)Time to the market -0.001(0.001) 0.000(0.001) -0.002*(0.001)Severity of shocks Asset loss -0.019***(0.002) -0.046(0.122) -0.017***(0.002)Income loss -0.003**(0.001) -0.001(0.001) -0.004*(0.002)
The Impact of Remittances on Asset Growth by Welfare Status in Thailand
Blinder–Oaxaca decomposition of growth difference by remittance status (%)
Results & discussions
18Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
Vietnam Thailand
Endowments Coefficients Total Endowments Coefficients Total
Human & social capital -0.42 22.57 19.43 0.80 -42.26 -41.46
Natural & physical capital 0.74 40.93 41.67 -1.25 9.85 8.60
Geographic & locational variables 1.93 -7.26 -5.33 1.49 21.15 22.64
Total 2.25 56.24 58.49 1.03 -11.26 -10.23
Summary of decomposition results
Total growth differential (E + C + U) 9.21 -7.25
Endowment & structural effect (C + E) 58.49 10.23
Unexplained effect (U) -40 -2.98
Coefficients as of attributable difference [C/(C + E)] 96 110.12
For Vietnam
Remittances (not migration) has significant impact in explaining asset
accumulation overtime
Structural poor migrant households with remittances experience higher
growth in asset
Remittances increase rural wellbeing by increasing productivity and
endowment effect to some extent
Shocks, limited accessibility of infrastructure facilities & ethnicity hinder
asset accumulation overtime
19Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
Results & discussions
For Thailand
Migration & remittances hinder rural households to accumulate assets &
escape from poverty
Migration & remittances decrease asset accumulation for structural poor
households,
It offsets the tendency of poor households to climb out of poverty
& catch-up to their better-off neighbors.
Shocks, limited accessibility of infrastructure facilities & gender offset
the tendency toward convergence
20Background Objectives Conceptual framework Data Econometric Approach Conclusions
Results & discussions
1. Poor households are more likely to migrate & receive remittances, but
tend to have low quality and return employment
2. Rapid & continuing out-migration of younger household members
increase labor constraints
3. Households receiving remittances tend to be structurally poor
households because they spend remittances for consumption purpose
Conclusions
21Background Objectives Conceptual framework Data Econometric Approach Conclusions
Policy recommendations
1. Migration & remittances increase productivity.
2. However, not all migration decisions lead to the expected success
3. High rates out-migration of young households members can result in
i. a decline in production & productivity in rural areas,
ii. bad employment in urban areas
Inequality in rural areas may continue growing
Conclusions
22Background Objectives Conceptual framework Data Econometric Approach Conclusions
Thank you for your attention
23
24
Variable Per capita inc. Annual per capita income (US$ PPP at 2005 prices) 1428.68
(861.35)Household & social capital HH-size Total household size 4.38(1.76)Children Proportion of children in household less than 15 0.24(0.22)Elderly Proportion of elderly in household above 60 0.10(0.23)Age Age of the household head 47.40(15.64)Mean edu. Average years of schooling of adult members 7.86(3.42)Primary Proportion of adult completed primary school 0.26(0.24)High school Proportion of adult completed high school 0.20(0.22)Professional Proportion of adult completed professional education 0.55(0.27)Gender Gender of the head (male headed =1, female headed =0) 0.77Ethnic Major ethnic Kinh & Hoa (=1), others(=0) 0.79Membership Any household member involved in political or social organization
(yes=1, no=0)0.87
Off-farm Participated in off-farm activities (yes=1, no=0) 0.52Self emp. Own small & medium scale enterprise (yes=1, no=0) 0.25
Description & Summary Statistics of Panel Variables (N = 6318)
25
Description & Summary Statistics of Panel Variables (N = 6318)Physical & natural capital measured in US$ PPP at 2005 prices in hundreds Agric. tools Value of agricultural tools owned 4.08(10.37)Transp. tools Value of transportation tools owned 10.70(29.54)L& Land size owned, in hectares 0.78(1.12)Livestock Value of livestock owned 8.11(25.88)Own house Value of house owned 102.22(158.17)House utilities Value of house utilities owned 10.08(19.66)Asset shock severity Low Experienced less sever (yes=1, no=1) 0.01Medium Experienced medium sever (yes=1, no=1) 0.09Highly Experienced highly sever (yes=1, no=1) 0.27Income shock severity Low Households experienced less sever (yes=1, no=0) 0.17Medium Households experienced medium sever (yes=1, no=0) 0.17Highly Households experienced highly sever (yes=1, no=0) 0.47
26
Description & Summary Statistics of Panel Variables (N = 6318)
Coping measures in responses to shocks Forest extrac. Depend on forest extraction (yes=1, no=0) 0.14Diversify agric. Diversifying agricultural profile (yes=1, no=0) 0.06Drawing assets Drawing drawn assets (yes=1, no=0) 0.13Lending informal Lent money from informal (yes=1, no=0) 0.17Public transfer Participated in public transfer (yes=1, no=0) 0.12Reduce cons. Reduced consumption (number of meals) (yes=1, no=0) 0.67Geographical capital at village level Paved road The village has paved road (yes=1, no=0) 0.55Mountainous The village is located in mountainous (yes=1, no=0) 0.48Main transp. Main transportation of the village is bus or motorcycle (yes=1,
no=0)0.58
Violence The village experienced violence (yes=1, no=0) 0.17Epidemics The village experienced epidemics (yes=1, no=0) 0.11Water supply The proportion of households with public water supply in the village 0.23Irrigated Total irrigated land in the village 13.01(24.21)No. of enterp. Number of enterprises who have more than 9 employees 0.17(1.64)HHs elect. The proportion of households with electricity in the village 92.31(22.65)HHs sanit. The proportion of households with sanitation in the village 18.20(31.53)Time to market Time to reach nearest market in minute 22.57(24.37)Time to bank Time to nearest bank market in minute 35.69(31.19)
20082007 Poor Non-poorPoor Twice poor 14 Rising from poverty 14 Stochastically poor 57 Stochastically mobile 65 Structurally poor 43 Structurally mobile 35Non-poor Declining into poverty 5 Twice nonpoor 67 Stochastically mobile 76 Structurally nonpoor 100 Structurally mobile 24 2010 2008 Poor Non-poorPoor Twice poor 10 Rising from poverty 9 Stochastically poor 55 Stochastically mobile 75 Structurally poor 45 Structurally mobile 25Non-poor Declining into poverty 7 Twice nonpoor 74 Stochastically mobile 50 Structurally nonpoor 100 Structurally mobile 50
27
28
Vietnam Thailand Migrant (38) Non-migrant
(62)Difference
TestMigrant (45) Non-migrant
(55)Difference
TestMonthly structural income 127(80) 115(84) *** 165(375 172(133) HHsize 4.46(1.76) 4.33(1.76) *** 4.12 (1.77) 3.97(1.71) ***Children 0.19(0.19) 0.28(0.22) *** 0.19(0.16) 0.21(0.18) ***Elderly 0.07(0.17) 0.12(0.25) *** 0.07(0.13) 0.11(0.22) ***Mean edu. 8.81(3.04) 7.27(3.51) *** 6.45(2.42) 5.90(2.45) ***Gender 0.78 0.76 * 0.72 0.73 Ethnic 0.86 0.75 *** - - Membership 0.91 0.85 *** 0.15 0.15
Results & discussions
29Background Objectives Conceptual framework Data Econometric Approach Results Conclusions
Structural income & poverty by migration status
2007 2010 Change
t-test
Annual
growth
2007 2010 Change t-
test
Annual
growth
Migrant Non-migrant
For Vietnam
Annual structural
income per capita
1332 1692 360*** 7.0% 1248 1476 228*** 5.0%
Structural poor 0.19 0.09 -0.10*** 2.5% 0.24 0.15 -0.09*** 2.3%
For Thailand
Annual structural
income per capita
1568 2127 559*** 9.0% 1541 2305 764*** 12.0***
Structural poor 0.20 0.14 -0.06** 1.5% 0.19 0.11 -0.08*** 2.0%