Você está na página 1de 6

Remittances and Working Poverty

JEAN-LOUIS COMBES*, CHRISTIAN HUBERT EBEKE**, MATHILDE MAUREL


& THIERRY URBAIN YOGO*
*Centre dEtudes et de Recherches sur le Dveloppement International (CERDI), Universit dAuvergne, Clermont-Ferrand,
France, **International Monetary Fund, Washington DC, USA, Centre dEconomie de la Sorbonne, Paris, France

Online Appendix
Table A1. List of countries and number of observations in the sample
Country
Albania
Argentina
Armenia
Azerbaijan
Burundi
Benin
Burkina Faso
Bangladesh
Bosnia and Herzegovina
Belarus
Belize
Bolivia
Brazil
Botswana
Central African Rep.
Chile
Cote d'Ivoire
Cameroon
Congo, Rep.
Colombia
Cape Verde
Costa Rica
Djibouti
Dominican Rep.
Algeria
Ecuador
Egypt
Ethiopia
Fiji
Gabon
Georgia
Ghana
Guinea
Gambia

Obs.

Country

Obs.

5
7
7
3
1
1
3
4
3
8
2
4
15
1
1
4
5
2
1
7
1
10
1
8
1
6
4
3
1
1
4
3
3
1

Liberia
Sri Lanka
Lesotho
Morocco
Moldova
Madagascar
Maldives
Mexico
Macedonia FYR
Mali
Mongolia
Mozambique
Mauritania
Malawi
Malaysia
Namibia
Niger
Nigeria
Nicaragua
Nepal
Pakistan
Panama
Peru
Philippines
Papua New Guinea
Paraguay
Russia
Rwanda
Senegal
Solomon Islands
Sierra Leone
El Salvador
Serbia
Suriname

1
4
3
3
6
5
2
10
5
3
1
3
2
1
5
1
4
3
4
2
5
10
6
6
1
7
6
2
4
1
1
9
1
1
(continued )

Table A1. (Continued)


Country
Guinea-Bissau
Guatemala
Guyana
Honduras
Haiti
Indonesia
India
Iran
Iraq
Jamaica
Jordan
Kazakhstan
Kenya
Kyrgyz
Cambodia
Lao PDR

Obs.
2
3
2
7
1
5
2
3
1
5
4
4
4
4
3
4

Country

Obs.

Swaziland
Syria
Togo
Thailand
Tajikistan
Tunisia
Turkey
Tanzania
Uganda
Ukraine
Uruguay
Venezuela
Vietnam
Yemen
South Africa
Zambia

2
1
1
7
2
2
4
2
4
5
2
5
4
2
4
2

Table A2. Descriptive statistics


Variable

Obs.

Mean

Std dev.

Min.

Max.

Percentage of working poor


Remittance-to-GDP ratio
ln (remittances per capita)
Remittance unpredictability
ln real GDP per capita in host countries
ln oil rents in neighbouring countries
ln real GDP per capita
ln (1+migration stock)
Migration instrumental variable, ln
Exports (% GDP)
FDI (% GDP)
Aid (% GNI)
Gini index

406
3176
3265
2308
4247
5617
5944
6348
6348
5296
4238
5041
1950

36.2
4.9
2.6
16.0
9.5
24.4
7.7
11.7
7.2
31.0
3.0
8.1
42.54

29.9
8.9
2.3
11.7
0.9
3.0
1.0
2.4
4.6
19.8
6.3
11.8
9.71

0
0
6.5
0
5.9
6
5
0
0
0.2
82.9
2.8
21.6

98.4
96.9
7.5
49.9
10.8
28.0
10.4
16.4
14.9
166.4
112.8
242.3
74.33

Table A3. Testing for attrition bias in the sample

Dependent variable: prevalence of working poor


Selection indicator (forward value)
Selection indicator (lagged value)
Remittance-to-GDP
Observations
Joint significance test of selection variables, P-value
Number of countries

(1)

(2)

(3)

OLS-FE

IV-FE

IV-FE

0.636
(0.842)
0.775
(0.745)
0.764**
(2.349)
333
0.301
95

0.209
(0.187)
1.383
(1.230)
1.978***
(4.607)
310
0.433
72

0.198
(0.131)
1.962
(1.276)
3.135***
(3.427)
310
0.440
72

Notes: T-statistics in parentheses. The models estimated in columns 1 and 2 include the full set of control
variables. In column 2, remittance inflows are instrumented by the GDP per capita in migrant host countries
while in column 3, the instrument used is the log of oil rents in neighboring countries. ***p < 0.01, **p < 0.05,
*p < 0.1.
The procedure (Semykina & Wooldridge, 2010; Verbeek & Nijman, 1992) consists in adding time-varying
functions of selection indicators as explanatory variables, and obtaining simple t or joint Wald tests. If we call
Si,t the indicator of selection which takes 1 when the dependent variable is observed at each year t and 0 otherwise,
we can add Si,t1, and Si,t+1 in the model for the working poor and test their joint significance. Under the null
hypothesis of no attrition bias, the coefficients of these variables (1 and 2 ) should not be statistically different
from zero. More formally, the equation to be estimated is:

wi;t 1 Si;t1 2 Si;t1 1 Ri;t X0 i;t Mi;t i i;t


Regardless of the specifications, the coefficients associated with the lagged and forward selection
dummy variables are not individually and jointly significant.

i,j

Common language,

ln (real GDP per capita),

1960
0.899***
(28.94)
4.417***
(24.30)
0.506***
(8.165)
4.160***
(21.96)
0.245***
(14.07)
0.286***
(16.42)
6.053***
(17.90)
9,409
9,409
0.284

0.954***
(100.4)
4.528***
(71.04)
0.646***
(32.04)
4.325***
(62.87)
0.130***
(26.71)
0.397***
(81.40)
6.288***
(62.34)
108,815
35,344
0.260

(2)

19602000

(1)

0.857***
(31.49)
4.708***
(27.46)
0.690***
(12.58)
4.364***
(24.66)
0.232***
(16.17)
0.365***
(25.46)
5.087***
(17.56)
13,689
13,689
0.262

1970

(3)

0.868***
(38.49)
4.740***
(30.53)
0.667***
(14.35)
4.637***
(29.11)
0.159***
(13.38)
0.361***
(30.34)
5.487***
(22.63)
19,044
19,044
0.249

1980

(4)

0.959***
(54.74)
4.506***
(37.43)
0.663***
(17.35)
4.339***
(32.24)
0.126***
(13.81)
0.382***
(41.78)
6.358***
(33.87)
31,329
31,329
0.256

1990

(5)

1.051***
(62.40)
4.410***
(37.69)
0.631***
(16.99)
4.058***
(30.96)
0.0806***
(9.696)
0.479***
(57.55)
6.919***
(39.09)
35,344
35,344
0.274

2000

(6)

Notes: The models are estimated using OLS on pooled data on global bilateral migration. Bilateral data on migration stocks are drawn from the World Global Bilateral
Migration Database by the World Bank. In column (1), the model is estimated using the full sample of countries. In columns 26, the model is estimated for each
corresponding year matching the data frequency in the World Global Bilateral Migration Database. Columns 26 have been used to generate the time-varying instrument for
migration calculated as the exponential of predicted dependent variable minus 1 for each year (1960, 1970, , 2000) and then averaged over each country i at each year t. Tstatistics are in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.

Observations
Country pairs
R2

Intercept

i,j

i,j

ln (real GDP per capita),

Colonial relationship,

i,j

Contiguity dummy,

ln (bilateral distance),

Dependent variable:
ln (1+stock of migrants from i living in j)

Table A4. Gravity model for bilateral migration

Description of the Control Variables and Expected Signs in the Baseline Regressions
GDP per capita refers to the level of real GDP per capita. A negative correlation between the level of
development and the prevalence of working poor is expected for two main reasons. First, poverty itself
is strongly and negatively correlated with mean income. Second, an increase in income per capita is
associated with higher education attainment, health stock and better opportunities, which increase the
reservation wage and reduce the prevalence of working poverty.
Stock of migrants abroad is an indicator of pressure on the labour market in the remittance receiving
country, but could also help control for incentive effects toward emigration, which can exert an impact
on the reservation wage, or even on the propensity of individuals to supply labour domestically. The
inclusion of this variable ensures that the effect of remittances is not confounded with the effect due to
migration. Migration changes both the country distribution of worker characteristics and the distribution of domestic wages. Hanson (2007) showed that in high migration Mexican states labour force
participation decreases, and that average wages in those states are between 6 per cent and 9 per cent
higher. We run subsequent regressions with and without the emigration variable to capture how strong
the effect of remittances on the prevalence of the working poor is, even when migration is taken into
account.
Exports-to-GDP ratio has an ambiguous impact. On the one hand, it should decrease the prevalence
of the working poor by increasing labour demand in the tradable sector and hence the associated wage.
International competition induced by trade openness leads to productivity gains in the tradable sector,
and moves the level of wages above their previous value. On the other hand, a high level of trade
openness means a higher exposure to external risks, which could translate into a lower reservation
wage and therefore increase the prevalence of working poverty. Finally, the more open countries
exporting natural resources may exhibit higher volatility, and Dutch Disease type effects, which hurt
the most vulnerable and lead more people to supply more labour. On the structural side, when
openness is the exporting highly capital intensive goods which requires specific types of skills and
human capital, low-skill workers may find it hard to supply labour, thus increasing the likelihood of
selling low wages.
Foreign direct investment net inflows-to-GDP controls for the fact that foreign direct investment
inflows increase domestic demand for labour, and hence reduce the prevalence of the working poor.
Chen, Ge, and Lai (2011) showed that FDI inflows increase inter-enterprise wage inequality in China.
Their results highlight the fact that FDI leads to a negative spillover in terms of the wage in domestic
firms. This arises when competition from multinationals may reduce the market share of local firms,
which drives such firms under the minimum efficiency or even crowds them out. Nunnenkamp,
Schweickert, and Wiebelt (2007) showed that even if FDI inflows enhance economic growth and
reduce poverty, they also widen income disparities between urban and rural areas in Bolivia. This
suggests that FDI inflows can exert a pro-poor effect in receiving countries, but at the same time they
can induce regional disparities.
Foreign aid can directly affect the share of the working poor through its pro-poor effects. These
effects may arise from either the labour supply or the labour demand sides. When foreign aid is used to
finance labour-intensive projects, there is an upward pressure on the labour demand which pushes up
wages, even for the poorest. On the supply side, foreign aid granted in the form of cash transfers or the
development of human capital and infrastructure could affect labour supply decisions and raise the
reservation wage. Finally, the Dutch Disease effects of foreign aid, which have been demonstrated by
recent papers (Rajan & Subramanian, 2011), could be a transmission channel if the poor are working
in the non-tradable sector. The net effect on working poverty is ambiguous.
Income inequality (approximated by the Gine coefficient of income distribution) is introduced in the
regression to ensure that the effect of remittances on working poverty is net of the effect of remittances
on income inequality. We expect a positive effect of income inequality through an increase in labour
supply (Blundell, Pistaferri, & Saporta-Eksten, 2012). In fact poor households respond to the growing
income gap by increasing their labour supply. This generates a downward pressure on wages and
raises the level of working poverty.

References
Blundell, R., Pistaferri, L., & Saporta-Eksten, I. (2012). Consumption inequality and family labor supply (Working Paper
18445). Cambridge, MA: NBER.
Chen, Z., Ge, Y., & Lai, H. (2011). Foreign direct investment and wage inequality: Evidence from China. World Development,
39(8), 13221332.
Hanson, G. H. (2007). Emigration, remittances, and labor force participation in Mexico. Integration and Trade Journal, 27,
73103.
Nunnenkamp, P., Schweickert, R., & Wiebelt, M. (2007). Distributional effects of FDI: How the interaction of FDI and economic
policy affects poor households in Bolivia. Development Policy Review, 25, 429450.
Rajan, R. G., & Subramanian, A. (2011). Aid, Dutch disease, and manufacturing growth. Journal of Development Economics,
94(1), 106118.
Semykina, A., & Wooldridge, J. (2010). Estimating panel data models in the presence of endogeneity and selection. Journal of
Econometrics, 157, 375380.
Verbeek, M., & Nijman, T. (1992). Testing for selectivity bias in panel data models. International Economic Review, 33,
681703.

Você também pode gostar