alexa Workers’ Remittances in Central and Eastern Europe (1993-2006): A Comparison to Latin America, and the Middle East | Open Access Journals
ISSN: 2162-6359
International Journal of Economics & Management Sciences
Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.
Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on
Medical, Pharma, Engineering, Science, Technology and Business

Workers’ Remittances in Central and Eastern Europe (1993-2006): A Comparison to Latin America, and the Middle East

Didar Erdinç1* and Ligia Dorobantu2

1Associate Professor of Economics, American University in Bulgaria, Bulgaria

2Senior Student, Economics, American University in Bulgaria, Bulgaria

*Corresponding Author:
Didar Erdinç
Associate Professor of Economics
American University in Bulgaria, Bulgaria
Tel: +359-888486
E-mail: [email protected]

Received date: June 18, 2014; Accepted date: December 01, 2014; Published date: December 10, 2014

Citation: Erdinç D, Dorobantu L (2014) Workers’ Remittances in Central and Eastern Europe (1993-2006): A Comparison to Latin America, and the Middle East. Int J Econ Manag Sci 3:211. doi:10.4172/2162-6359.1000211

Copyright: © 2014 Erdinç D, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Visit for more related articles at International Journal of Economics & Management Sciences

Abstract

Despite the growing importance of workers’ remittances in total capital flows, the relationship between growth and remittances has not been adequately studied empirically in the context of transitional economies, especially in a comparative framework with other recipient countries. This is surprising because migrants from the Central and Eastern European Economies (CEECs) contribute significantly to their home economies through remittances, influencing investment and consumption patterns. This paper examines the impact of workers’ remittances on growth, investment and consumption in a number of CEECs in comparison to Latin America and the Middle East-North Africa- Turkey (MENAT) in a panel data framework. Using annual data ranging from 1993-2006, it is shown that as compared to Latin America, both investment and consumption are positively affected by the amount of remittances sent by workers’ to their home countries in the CEECs after controlling for several important determinants of growth such as openness to trade, inflation, real interest, credit-GDP ratio as a measure of financial deepening. In the CEECs, as in the Middle East, the workers’ remittances affect growth both through consumption and investment while the latter effect is stronger. By contrast, in Latin America, remittances mainly impact consumption rather than investment, even having a negative impact on growth. To gauge these differential effects, we use fixed and random effects estimation as well as GMM strategy to account for country-specific heterogeneity and to control for possible endogeneity among repressors. Our findings also suggest that workers’ remittances could be a significant impetus for growth, working through the investment channel, and their significance conditional on credit in investment equations suggest that they can help overcome the liquidity constraints by providing an alternative to formal channels of financing.

Keywords

Growth; GMM; Investment; Panel estimation; Remittances

Introduction

In the past two decades, since the collapse of communism, migrant workers’ remittances have become an important source of external finance for transitional economies in Central and Eastern Europe (CEECs), second only to the FDI flows. In certain countries, such as Albania, Croatia, Macedonia and Moldova, their level has been a few times larger than that of foreign direct investment (FDI), sometimes reaching a significant part of GDP (Figures 1 and 2). Despite the proliferation of research work that explore the potentially beneficial impact of remittances in developing countries, and the growing significance of remittances for the CEECs in the past decades, only a limited number of papers assessed their macroeconomic effects in this region, especially in a comparative perspective with other recipient countries in Latin America, the Middle East, North Africa and Turkey (MENAT). Most work so far has been qualitative and at the cross-national level, there is, to the best of our knowledge, only one empirical study dedicated to the macroeconomic study of remittances in the CEECs. Piracha and Ledesma [1] analyze the impact of remittances on investment, and consumption in several countries of Central and Eastern Europe but their paper excludes the possible role of remittances on growth, especially after controlling for indicators of financial development.

economics-and-management-GDP-CEECs

Figure 1: Remittances as % of GDP in the CEECs.

economics-and-management-FDI-CEECs

Figure 2: Remittances as % of FDI in the CEECs.

On the other hand, the results of macroeconomic analyses of remittances in developing economies point in different directions: There are two contradictory views regarding the impact of remittances on the macro economy despite the encouraging results of several micro-studies.1 One view stresses that remittances are primarily spent on consumption and residential investment, having little or no effect on growth and capital formation. Remittances can even hurt the growth process by discouraging work effort, thereby reducing labor supply. Chami et al. [2], for instance, find that remittances are compensatory in nature and create moral hazard problems, thus negatively affecting economic activity and growth.

Yet, another strand of literature argues that remittances are potentially productive when directed towards investment in physical and human capital, both of which are important determinants of growth. Rapoport and Docquier [3] suggest that remittances can have a significant contribution to education, thus having a positive effect on growth through the human capital channel. Glystos [4] analyzes the impact of remittances on consumption, investment, output and imports in five countries and finds that remittances enhance growth, but there are also cases when remittances decrease growth or intensify recessions. Ziesemer [5] finds that remittances have a positive effect on growth and that this effect is more significant for poorer countries. In a recent paper, Acosta et al. [6], in a macro-panel framework, argue that remittances have so far boosted growth and reduced poverty and inequality in Latin America. Giuliano and Ruiz-Arranz [7], in a large sample study of developing countries in a macro setting, find that remittances not only enhance growth but also relieve borrowing constraints in countries with shallow financial markets.

Our paper contributes to the debate on the possible impact of workers’ remittances on the macroeconomy in two important ways: First, we analyze the effect of remittances on consumption, investment and growth in a comparative context by dividing the sample into three regions: the CEECs (11 countries), and two other sets of top remittance – receiving regions, broadly grouped as Latin America (16 countries including those in the Caribbean region) and MENAT (7 countries) over the 1993-2006 period. Our objective is to analyze, using an paneleconometric approach, whether remittances have a strong positive effect on growth in the CEECs in comparison to other regions and whether this effect works mainly though investment. To this end, we estimate three equations on the determinants of consumption, investment and growth for each region to assess the possible differential impact of remittances in the CEECs relative to the other two regions. Our strategy is motivated by the observation that pooling a sample of developing countries to study the impact of remittances may be misleading as there may be substantial amount of variation across regions regarding the effect of remittances on growth, consumption and investment invariant country-specific heterogeneity for each region. In the meantime, we control for the indicators of financial development in all specifications, otherwise standard, to explore how financial sector development affects the ability of a country to benefit from remittances and whether this effect differs in these three remittance receiving regions. Our paper closely follows, in this respect that of Giuliano and Ruiz–Arranz that examines the impact of remittances on economic growth for developing countries by looking at the way local financial sector development influences a country’s capacity to take advantage of remittances.2 We wish to study whether their results generalize to the CEECs in comparison to Latin America and the MENAT. Moreover, unlike Piracha and Ledesma, we perform dynamic Generalized Method Moments (GMM) estimation as in Arellano and Bond [8] and Arellano and Bover [9] for consumption and investment (also for growth equations), using internal instruments to control for simultaneity bias. We use data from the World Bank World Development Indicators.

Our key findings can be summarized as follows: Remittances impact growth in a significant manner via capital accumulation, and hence work through the investment link in both CEECs and the Middle East. Yet, remittances mainly finance consumption in Latin America, having a negative impact on growth (seemingly) in contrast to the results of Acosta et al., but this effect is insignificant.3 There is also evidence that in CEECs and the Middle East, remittances raise consumption and this effect is statistically significant. In growth regressions, conditional on investment and human capital, the impact of remittances on growth (i.e. total factor productivity) is significantly positive only in one of the specifications for CEECs and MENAT but insignificant and negative in Latin America. Hence, there is some evidence that remittances also affect total factor productivity in the CEECs and the Middle East but not in Latin America.

The rest of the paper is organized as follows: Next section presents the empirical analysis using fixed and random effects as well as GMM estimation methodology in the presence of country-specific heterogeneity within each region. First, we describe the variables used in each regression specification for consumption, investment and growth and their expected signs. Then, we proceed with the specification tests such as Hausman and Sargan tests after a brief discussion on the challenges associated with the potential endogeneity of remittances and growth in the presence of country-specific effects.

Significance of remittances in consumption, investment and growth regressions for CEECs in comparison to other regions is discussed based on the estimation results. Section 3 offers concluding remarks.

Empirical Analysis

Description of variables and their expected signs

We have performed panel data econometric estimation on 3 subsamples of remittance receiving countries: 11 countries in Central and Eastern Europe, 16 countries in Latin America and the Caribbean, and 7 countries in the Middle East and North Africa, including Turkey using annual data for 1993-2006. We have chosen the countries in these three regions based on their high level of remittances as percentage of their GDP levels. There are missing data in some countries for certain variables, thus the number of observations changes according to the variables included.

The remittance variable,REM, represents the sum of workers’ remittances and compensation of employees working abroad as it appears in the World Bank WDI database. In the investment regression, the dependent variable is gross fixed capital formation,INV. Other set of control variables, in addition to REM, include the following variables: We proxy the user cost of capital by the real interest rate, ri and financial depth by the credit provided by the banking sector as a measure of financial intermediation performed by the banking sector, CRED. We also control for the real GDP growth, GRWTH and foreign direct investment, FDI. We expect REM to have a positive effect on investment along with CRED, grwth and FDI while RI is expected to have a negative effect on investment. Our findings confirm these expectations with the exception of the impact of the real interest rate which is negative only in Latin America.

In the CONSUMPTION equation, the dependent variable is final consumption expenditure, cons. We include GDP per capita,GDPPC; the real interest rate,RI; inflation rate defined as the annual percentage change in the consumer price index, INFL and unemployment rate, UNEMP together with lagged consumption, LAGC and remittances,REM as explanatory variables. We expect gdppc and rem to have a positive effect on consumption while RI, INFL and UNEMP to have a negative effect. Later, we show that our expectations are confirmed in all regions regarding the signs of these coefficients in consumption equations.

For the GROWTH regressions, the real GDP growth, GRWTH, is our dependent variable. In addition to remittances, REMG (as a percentage to GDP), our set of control variables include lagged value of GDP per capita, GDPPC; inflation rate, INFL; openness to international trade, defined as the sum of exports and imports as percentage of GDP, to; log of secondary school enrollment a measure of human capital, ED; gross fixed capital formation as a percentage of GDP, invg; domestic credit provided by banking sector as a percentage of GDP as a measure of financial depth, credG. This last variable shows how much financial intermediation is performed by the banking sector, as a proxy for financial sector development.

The variables in the investment and consumption equations appear in level terms; for the growth equations we use ratios (to GDP). All variables are specified in natural logs (Tables 1 and 2).

Country N Mean St. Dev Minimum            Maximum
CEEC 139 4.25918 6.99135 0.01347 36.1957
Albania 14 16.7129 3.450217 11.85068 27.03427
Bulgaria 11 3.370482 3.464496 0.33215 8.596169
Croatia 14 3.135812 0.486956 2.109586 3.840063
Hungary 12 0.407833 0.099217 0.271827 0.585927
Macedonia 11 2.722273 0.926549 1.53771 3.967632
Moldova 12 6.39409 11.38547 0.057045 36.19567
Poland 13 0.851937 0.310054 0.491242 1.288331
Romania 13 0.805208 1.582761 0.025369 4.788317
Slovakia 14 0.408205 0.416863 0.0883 1.285743
Slovenia 14 1.043288 0.351805 0.410327 1.897722
Ukraine 11 0.330573 0.286596 0.013466 0.69076
Latin Am. 217 4.45215 5.46377 0.00206 27.0343
Argentina 14 0.093175 0.091155 0.020944 0.238721
Brazil 14 0.355574 0.10597 0.19459 0.538749
Bolivia 14 1.377308 1.100001 0.0698 3.580049
Chile 7 0.013466 0.005699 0.00206 0.0178
Colombia 14 1.926007 1.099072 0.72562 3.873527
Costa Rica 14 1.19353 0.643308 0.161023 2.30751
Dominican R. 14 9.214676 2.25619 6.666728 14.2484
Ecuador 14 4.458553 2.216555 1.347653 8.292747
El Salvador 14 13.17695 2.442688 10.50842 18.13598
Guatemala 14 4.802633 0.307414 2.114046 10.27479
Haiti 14 13.27629 8.354016 1.807672 27.03427
Jamaica 14 12.58535 3.914443 4.886311 18.29431
Mexico 14 1.763926 0.659136 0.977414 2.947156
Paraguay 14 3.441956 0.855236 1.38186 4.839211
Peru 14 1.310148 0.311694 0.829633 1.956705
Venezuela 14 0.025545 0.030958 0.002671 0.102194
MENA+T 98 3.62159 2.24411 0.21132 12.1601
Algeria 14 2.161311 0.56654 1.214186 3.279068
Egypt 14 4.903253 2.379421 2.856612 12.16008
Morocco 14 7.174604 1.450961 5.498064 9.619148
Pakistan 14 3.042074 1.179514 1.466159 4.972064
Syria 14 2.324092 1.33703 0.625514 5.285506
Tunisia 14 4.230652 0.684213 3.052924 5.090784
Turkey 14 1.51517 0.910313 0.211318 2.682912

Table 1: List of Countries and Worker Remittances as % of GDP, 1993-2006.

Country N Mean St. Dev Minimum Maximum
CEEC 128 142.268 219.248 0.58477 1119.751
Albania 13 595.5515 226.8642 337.1925 1119.751
Bulgaria 10 41.65461 42.32544 5.251664 130.1042
Croatia 13 135.0843 127.8403 38.18386 475.9405
Hunagry 11 7.104419 2.929812  3.16393 13.54922
Macedonia 10 212.8793 194.7942 16.5334 606.6013
Moldova 11 353.9157 242.288 3.859514 823.4058
Poland 12 27.56623 14.37447 11.34801 57.85574
Romania 12 11.42634 19.24993 1.316872 71.39011
Slovakia 13 18.13389 21.37042 0.584767 75.81374
Slovenia 13 105.7915 73.84001 13.07779 233.8931
Ukraine 10 11.65894 10.99594 1.151631 30.15873
Latin Am. 200 494.191 2675.98 -22596 15796.61
Argentina 13 4.4456 5.015705 0.266803 16.58585
Brazil 13 28.83625 29.41323 5.030623 96.51703
Bolivia 13 28.81531 97.86302 -122.189 320.9537
Chile 6 0.277452 0.106215 0.167301 0.4706024
Colombia 13 71.35888 44.37945 13.91532 175.0043
Costa Rica 13 31.65905 14.99965 5.712366 55.81979
Dominican R. 13 316.7139 221.6511 121.9166 997.9275
Ecuador 13 108.7562 42.0542 42.85232 183.6035
El Salvador 12 -608.566 7049.809 -22596.2 4844.193
Guatemala 13 695.4328 614.1406 67.92509 1675.966
Haiti 13 6063.658 5949.494 -2607.14 15796.61
Jamaica 13 277.4232 95.85931 150.8497 443.0122
Mexico 13 63.07807 27.05244 37.38187 116.7537
Paraguay 13 429.7657 519.1667 83.65019 2020
Peru 13 43.63352 2.12764 14.38045 88.67517
Venezuela 13 0.984063 1.356571 0.183234 5.005073
MENA+T 91 403621 2198955 8.67925 1.40E+07
Algeria 13 2811712 5379189 60.19226 1.40E+07
Egypt 13 489.0336 341.9813 93.32912 1247.262
Morocco 13 12048.57 23428.64 156.2687 73080.63
Pakistan 13 332.0905 158.0628 139.2661 742.3221
Syria 13 251.2095 135.012 66.66667 555.625
Tunisia 13 186.8769 67.43648 79.3739 308.8054
Turkey 13 324.8493 222.2305 8.679245 578.4163

Table 2: Workers’ Remittances as % of FDI, 1993-2006.

Growth is expected to be positively affected by remittances, openness, human capital, investment and financial development but negatively by the lagged value of real GDP (in the case of conditional convergence in the regional sample), and inflation (a measure of uncertainty and instability in economic environment or policies). We find evidence for conditional convergence only for CEECs and Latin America but not for MENAT. Similarly, inflation affects growth process negatively only in CEECs and Latin America but positively in MENAT. But this effect is insignificant in most specifications. Openness to trade is highly significant with a positive coefficient in Latin America, and to a lesser extent in CEECs but insignificant in MENAT despite having the correct sign. Surprisingly, human capital enters with a negative sign into the growth equations for CEECs although it is insignificant in all specifications as in Latin America. In MENAT, it has a positive effect on growth but only significant in one specification. The impact of remittances, after controlling for investment in physical and human capital is significantly positive (under GMM specification) only for CEECs and MENAT but not for Latin America. In addition, for Latin America, remittances variable carries a negative sign, but is insignificant (Tables 5a-5c).

Regression Analysis: Fixed versus Random Effects Estimation

We run three regional regressions for CEECs, Latin America and the Middle East for three equations pertaining to consumption, investment and growth. Based on the F-test, we reject the hypothesis of pooled estimation in favor of fixed effect (FE) estimation. Pooled OLS estimates are biased in the presence of country specific heterogeneity and fixed effect estimation should be favored under these conditions [10]. This result implies that a significant amount of country heterogeneity is present in all three regions for these dependent variables during 1993-2006. In order to control for common shocks, or external disturbances (like changes in world interest rates), we estimated all fixed effect equations with common time dummies captured by λt but later dropped them due to their joint insignificance. We also performed Hausman specification tests for each equation and region: For CEECs and MENAT, the test favored fixed effects (FE) over random effects (RE) for the consumption equation. For Latin America, random effects estimation was the preferred choice. For the investment equation, Hausman test favored fixed effects for Latin America and MENAT but random effects for the CEECs. For growth equations, random effects, again, is favored for CEECs but fixed effects for Latin America and the Middle East. We report the results for both FE and RE in the Tables 3-5.

Dep. variable:  ln(cons) LSDV1
w/out d1
RE FE GMM
ln(gdppc) .6168115***(2.88) 0.015286 (0.07) 1.573008 *** (5.86) 0.672628(0.8)
ln(rem) .0782774***(3.24) 0. 0071502 (0.11) .1268257 *** (3.67) .0642453***(2.68)
ln(ri) -0.0141885(-0.49) -.4152403*** (-2.99) 0.062049 (1.45) -0.03992(-1.11)
ln(infl) 0.0093833(0.2) -0.14214 (-1.00) .0297686 (0.43) -0.04011(-0.87)
ln(unemp) -.152085**(-2.14) -.6101889** (-2.29) -0.18053 (-1.60) -0.15216(-1.48)
lagc .7557432***-7.05      
constant 0.00719280 25.92848 (9.14) 8.188365 (3.58) -0.00609(-0.18)
# Obs. 42 43 43 34
R2 overall 0.9876 0.3626 0.1066  
F - test     0.0005  
BP LM test   0    
Hausman test     0  
Sargan test       0

Table 3a: Panel Estimation of the Consumption Equation for CEECs.

Dep. variable:  ln(cons) LSDV1
w/out d1
RE FE GMM
ln(gdppc) 1.158215***(4.73) 1.62901***(7.56) 1.697104***(6.55) 1.910376***(6.85)
ln(rem) -0.0300462(-1.25) .0474396*(1.93) 0.0397069(1.53) .0760231***(2.6)
ln(ri) -.0402092**(-2.26) -.0383525*(-1.75) -.0403772*(-1.82) -.0449652**-0.027
ln(infl) -.0903968***(-4.72) -.1310658***(-5.97) -.1316398*** (-5.89) -.0854351***( -4.96)  
ln(unemp) -.1593294***(-2.65) -0.0756426(-1.10) -0.0626717(-0.87) -0.0797975(-1.60)
lagc .445042***(6.74)      
constant 5.176751(2.54)   10.47435(5.33) -0.0267487(-3.66)
# Obs. 113 123 123 85
R2 overall 0.9952 0.4013 0.3944  
F - test     0  
BP LM test   0    
Hausman test   0.9408    
Sargan test       0.0007

Table 3b: Panel Estimation of the Consumption Equation for Latin America.

Dep. variable:  ln(cons) LSDV1
w/out d1
RE FE GMM
ln(gdppc) -0.5163297(-1.12) 2.850368***(6.10)  2.286184***(4.42) 2.059083** (1.99) 
ln(rem) 0.0614286 (1.38) .2346272*** (2.8) 0.0058346 (0.06) .0706712** (2.01)
ln(ri) -0.061375 (-0.83) .4334042** (2.04) 0.0246157 (0.17) -0.0862492(-1.30)
ln(infl) -0.0135542 (-0.37) 0.1473951 (1.56) -0.0625405 (-0.83) -0.0170374(-0.58)
ln(unemp) -0.0720066 (-0.80) -.5451023*** (-2.46) -0.0657156 (-0.38) -.1992247** (-2.16)  
lagc .9911579*** (7.05)      
constant 3.264694 (1.37)   2.977991 (1.27) -0.0590278(-2.64)
# Obs. 23 25 25 15
R2 overall 0.9942 0.9029 0.4808  
F - test     0.0002  
BP LM test   0.7587    
Hausman test     0  
Sargan test       1

Table 3c: Panel Estimation of the Consumption Equation for MENAT.

Dep.variable: ln(inv) LSDV1
w/out d1
RE FE GMM GMM
ln(rem) 0.0264615(1.5) 0.005024(0.21) .0323219  (1.18) 0.015158(0.95) .0249302*(1.65)
ln(cred) .3068385***(5.78) .7921998***(19.72) .7200371***(13.43) .3528811***(7.71) .3292005*** (7.81)
ln(ri) 0.0369406 (1.6) .0628744* (1.76) .0588078  (1.65) .0361845* (1.93) 0.026296(1.64)
ln(fdi) 0.0213651 (1.26) .0808486*** (3.22) .0749797*** (2.98) 0.000692 (0.04)  
ln(grw) .0955764*** (4.85) .0585144* (1.95) .0518667* (1.74) .0804809*** (5.49) .0750993***(5.04)
laginv .6018088***(10.19)        
constant  .6623902  (1.04)   3.449413  (3.63) 0.01045(1.45) 0.006378(1.09)
# Obs. 86   87 70 76
 F test     0    
BP LM test   0      
Hausman test   0.8078      
R2 overall 0.9961 0.9682 0.9656    
Sargan test       0.0805 0.0841

Table 4a: Panel Estimation of the Investment Equation for CEECs.

Dep.variable: ln(inv) LSDV1
w/out d1
RE FE GMM
ln(rem) 0.0254038(1.34) -0.0037942(-0.16) 0.0252462(0.95) 0.024949(0.76)
ln(cred) .08322*(1.89) .4654617***(9.34) .3512484*** (6.16) .2853827***(3.9)
ln(ri) -0.0028384(-0.17) -.0553971**(-2.08) -.0450623* (-1.73) 0.0299624(1.6)
ln(fdi) .0521318***(2.91) .1722576***(6.72) .1684958 ***(6.78) .0526528***   (2.92)
ln(grw) .0808074***(6.52) .054892***(2.81) .0481745 ***(2.55) .0568764***(4.65)
laginv .7237957***(13.9)      
constant 3.038745(3.23)   10.50455 (10.38) 0.0057656(0.413)
# Obs. 131 142 142 100
 F test 0.9961 0.9209 0.9222  
BP LM test     0  
Hausman test   0    
R2 overall     0.0002  
Sargan test       0.0025

Table 4b: Panel Estimation of the Investment Equation for Latin America.

Dep.variable: ln(inv) LSDV1
w/out d1
RE FE GMM
ln(rem) 0.1018792(1.7) -0.0462207(-0.58) 0.0660506(0.83) .0946408**(2.19)
ln(cred) 0.1348051(0.75) .4570562***(4.91) .5980138*** (5.68) .7843866*** (3.08)
ln(ri) .0469719* (1.97) 0.0063784 (0.17) 0.0311124 (0.93) .0457568***(2.72)
ln(fdi) .0384277* (1.82) .1242065*** (3.02) 0.0473043 (1.58) .0600953***(2.67)
ln(grw) 0.052867(1.39) -0.0620218(-0.80) 0.0100841(0.19) 0.0426176(1.5)
laginv .668268***(3.36)      
constant 1.518562(0.65)   6.150591(2.18) -0.0391014(-3.04)
# Obs. 31 33 33 21
 F test 0.9791 0.8681 0.833  
BP LM test     0  
Hausman test   0.0354    
R2 overall        
Sargan test       1

Table 4c: Panel Estimation of the Investment Equation for MENAT.

Dep.variableln(grw) LSDV1 w/out d1 RE FE FE - IV (IV:lagremg) FE - IV (IV:lagremg) GMM Endogenous Remittances
laggdppc -1.318034(-1.47) -.351842***(-2.94) -1.318034(-1.47) -1.383059(-1.46) -1.673303** (-2.56) -11.69551***(-3.05)
lnremg -0.0244585(-0.20) -0.0583547(-0.99) -0.0244585(-0.20) 0.0735341(0.17) .2989416*** (2.9) 0.0587748(0.55)
lninvg 0.3595301(0.54) 1.238705**(2.38) 0.3595301(0.54) 0.2782602(0.37) 0.1847494(0.43) 2.107151**(2.03)
lncredg 0.7759712(1.27) 0.1405249(0.65) 0.7759712(1.27) 0.5452078(0.48) 0.274094(0.89) 0.119248(0.16)
lned -0.4452894(-0.16) 0.1513837(0.11) -0.4452894(-0.16) -0.2654467(-0.09)   -4.430261(0.275)
lnto 1.956926**(2.13) 0.3136361(1.16) 1.956926**(2.13) 2.003522**   (2.12) 0.7376033(1.12) -0.0106912(0.993)
lninfl -0.0785364(-0.61) -0.035346(-0.36) -0.0785364(-0.61) -0.0840609(-0.64) -0.0453619(-0.55) .2829787*(1.81)
constant 1.084648(0.09) -2.073726(-0.35) 0.9177398(0.07) 1.412942(0.11) 9.603717(2.56) 0.6615423(2.71)
# Obs. 58 58 58 58 113 35
R2 overall 0.4293 0.315 0.1888 0.1699 0.0564  
F test     0.524      
BPLM test   0.0752        
Hausman test   0.3526        
Sargan test           1

Table 5a: Panel Estimation of the Growth Equation for CEECs.

 

Dep.variableln(grw) LSDV1 w/out d1 RE FE FE - IV (IV:lagremg) GMM Endogenous Remittances
laggdppc -3.632857 (-1.17) .020155 (0.04) -3.632857 (-1.17) -5.70606   (-1.30) -1.925991   (-0.25)
lnremg -.2827049 (-0.75) -.2488973 (-1.49) -.2827049     (-0.75) 1.462406   (0.79) -1.011484   (-1.28)
lninvg .166049 (0.11) -1.241062 (-1.43) .166049   (0.11) .6478027   (0.33) .5680088   (0.22)
lncredg -.7160751   (-0.76) -.6658217 (0.126) -.7160751   (-0.76) -2.03723   (-1.14)   -.3206416   (-0.19)
lned 1.529338   (0.80) .2118556 (0.29) 1.529338   (0.80) -2.485197 (-0.52)  -2.156925   (-0.70)
lnto 4.875535*** (3.11) 1.834339** (2.45) 4.875535*** (3.11) 1.380868 (0.34) 7.332959   (0.021)
lninfl -.3670645* (-1.72)    -.0748455 (-0.42) -.3670645* (-1.72)    -.328119   (-1.23)   -.069447   (-0.20)
constant -3.299439 (-1.90)   7.027886   (0.29) 55.17566 (0.95)  .32908   (1.23)
# Obs. 60 60 60 60 31
R2 overall 0.5894 0.2436 0.0130 0.0006   
F test     0.0088    
BPLM test   0.7157      
Hausman test          
Sargan test         1.0000

Table 5b: Panel Estimation of the Growth Equation for Latin America.

Dep.variableln(grw) LSDV1 w/out d1 RE FE FE FE - IV (IV:lagremg)
laggdppc 2.438879 (0.27) .5692951 (0.46) 2.438879 (0.27) 4.121234 (0.81) 2.086556 (0.23)
lnremg -1.597992 (-1.44) -.0148686 (-0.04) -1.597992 (-1.44) -1.703668** (-2.38) -1.528932 (-1.34)
lninvg -3.395252 (-1.17) -.811759 (-0.35) -3.395252 (-1.17) -4.077343 (-1.65) -3.408536 (-1.17)
lncredg .0329688 (0.02) -.1761047 (-0.34) .0329688 (0.02)   -.0020308 (-0.00)
lned 6.678835 (1.77) -.1446852 (-0.20) 6.678835 (1.77) 7.451272** (2.48) 6.472908 (1.67)
lnto 1.585579 (0.94) -.3187314 (-0.41) 1.585579 (0.94) 1.588365 (1.17) 1.612279 (0.95)
lninfl .0979344 (0.33) .1909196 (0.90) .0979344 (0.33)   .0856077 (0.29)
constant -39.18312 (-0.52) 2.079865 (0.58) -37.4792 (-0.52) -50.04385 (-1.31) -34.09443 (-0.46)
# Obs. 60 60 60 60 31
R2 overall 0.7354 0.3153 0.0179 0.0062 0.0186 
F test     0.2944    
BPLM test   0.5383      
Hausman test          
Sargan test          

Table 5c: Panel Estimation of the Growth Equation for MENAT.

image(1)

image(2)

image          (3)

where image and image

In the presence of endogenous remittances (remittances affect growth but growth or investment may also affect remittances), FE and RE estimates are inconsistent as they assume strictly exogenous regressors. Hence, we also estimate these equations using Arellano and Bond -Generalized Method of Moments (GMM) to address endogeneity problem and gauge the possible dynamic structure in the equations. GMM yields consistent estimates under these circumstances. There exists considerable difference in the estimated coefficients in these three cases- FE, RE and GMM in CEECs (consumption, and investment) and all growth regressions which suggest that a dynamic panel specification via GMM is most suitable in terms of generating consistent estimates.

Dynamic Panel Estimation: Generalized Method of Moments (GMM)

In order to account for the possible persistence in consumption and investment and their dynamic structure, we performed GMM estimation as in Arellano and Bond (1991).4 We reran the above regressions by using lagged values of consumption, investment and growth variables as instruments for dynamic equations in differences. Fixed or random effect models generate biased results in the presence of lagged dependent variables in the regression equations.5 FE model is appropriate for static models in which the regressors are correlated with the country-specific effects but FE requires strict exogeneity of the explanatory variables with respect to the random error term. Ideally, we should have conducted a Durbin-Wu-Hausman test to determine if some explanatory variables are endogenous. If they are, then FE and RE are both inconsistent. It is important then to use estimators that are consistent in the presence of endogenous regressors and country specific effects. GMM offers a robust solution to the endogeneity problem since it yields consistent estimates in the presence of endogenous regressors.6

Are remittances endogenous in growth, consumption and investment regressions? Does it depend on the income per capita in the recipient countries as a negative relation with poorer countries receiving more remittances? Theoretically, both the magnitude of remittances and the efficiency of financial markets as proxied by the credit variable should increase with higher growth rates. In this case, their effect on growth can be overestimated. Then if remittances depend on the level of income and if there is conditional convergence towards the steady state in per capita income, remittances can not be considered exogenous with respect to growth as traditionally assumed. At most, we may hope it is predetermined such that remittances may be influenced by random events in past growth rates but not by contemporaneous events. In the absence of good instruments, the endogeneity problem can be tackled with system GMM (SGMM) following Arellano and Bover. We also have tried to estimate growth equations in differences with using lagged values of endogenous variables as instruments. In GMM estimations, Sargan test of over-identifying restrictions also confirm the validity of the internal (lagged values of endogenous variables) used as instruments.

Do remittances have an impact on growth? To answer this question, we include several variables in the growth regression. The result is that, conditional on investment in physical and human capital as well as other variables that proxy financial development, the impact of remittances on growth is significant only in one of the specifications for CEECs and MENAT but almost negligible in Latin America. If remittances have an effect on growth, conditional on investment and human capital, then remittances work through channels that impact on total factor productivity or the Solow residual. We interpret this finding as supportive of the view that remittances affect growth via some of the control factors such as through their impact on capital formation or alternatively, on investment in physical capital.

Hence, we conjecture that remittances impact growth via capital accumulation, such that remittances work through the investment link. This implies that our estimation results should indicate that 1) investment impacts growth and 2) remittances impact on investment. Tables 3-5 reveal that this is indeed the case: Remittances strongly affect investment in CEECs, and MENAT even after controlling for the level of financial deepening through the credit variable. But this effect is not present for Latin America (negative and insignificant). On the other hand, remittances do have a positive and significant effect on consumption for all regions, which is especially strong for Latin America. Combining these results with our earlier findings on growth, we confirm that remittances work through investment (capital accumulation) in influencing growth and in regions (Latin America) where remittances mainly influence consumption, their impact on growth is also negligible. In such countries, remittances are likely to be devoted to non growth generating activities such as consumption and even may reduce labor supply and discourage growth. We also find such a negative (but insignificant) effect of remittances on growth in Latin America.

Do countries in each region converge in terms of real income? β1 of initial real income per capita in the growth regressions is the convergence coefficient, which have been included in almost all empirical growth studies in the 1990s. If the log of initial level of income is significant and negative, this implies conditional income convergence among the countries in the sample. We find evidence that there is strong conditional convergence in the growth equations for CEECs but not for Latin America and for MENAT (possibly reflecting greater heterogeneity of these samples). Moreover, although in Latin America, the coefficient has the right sign (but insignificant), in the MENAT group, it is positive (but insignificant).

Consistent with the previous literature, we find that financial development is conducive to growth via its impact on investment and growth in all regions. The coefficient of credit variable is consistently significant and positive in all specifications and regions. But it turns insignificant (even negative) in the presence of investment variable in growth equations displaying high level of multicollinearity with this variable. Moreover, credit is strongly significant in the investment equations in all regions but the coefficient is largest in MENAT, and larger in CEECs than in Latin America (sensitivity of investment to credit is highest in MENAT, and then CEECs). Combining these results, we can claim that remittances may alleviate credit constraints by providing an additional channel for investment financing in CEECs and MENAT more than Latin America where remittances finance mostly consumption.

Conclusions

This paper shows that pooling a sample of developing countries to study the impact of remittances may be misleading as there is substantial amount of variation across regions regarding the effect of remittances on growth and investment. While remittances affect positively investment, consumption, and growth (TFP) in the CEECs and the Middle East (MENAT), their main impact is on consumption in Latin America with a negative (albeit insignificant) impact on growth (reducing labor supply). This implies that remittances respond to profit opportunities at home (investment effect) and possibly alleviate credit constraints by financing start-ups. Consumption effect is strongest in Latin America, followed by the CEECs and the Middle East, but it discourages growth only in Latin America (but this effect is not statistically significant).

We also find that there is significant amount of conditional convergence in the CEECs and Latin America in terms of real per capita income but not in the Middle East. Consistent with the previous literature, we find that financial development is conducive to growth via its impact on investment in all regions. The coefficient of credit variable is consistently significant and positive in all specifications and regions but the coefficient is largest in MENAT, and larger in CEECs than in Latin America i.e. sensitivity of investment to credit is highest in MENAT, and then CEECs. Remittances may alleviate credit constraints by providing an additional channel for investment financing in the CEECs and MENAT (as it is significant even in the presence of the credit variable).

To summarize, remittances seem to finance both investment (profit-driven motive to remit) as well as consumption (compensatory transfers or altruistic motive to remit) in CEECs and MENAT where sensitivity to credit is larger (relieving financing constraints) than in Latin America where it finances mainly consumption rather than investment with remittances turning insignificant in investment equations. These interpretations are also in tune with our earlier findings on the growth effect of remittances in Latin America where remittances have a negative (but insignificant) effect on growth. It also confirms that when remittances affect growth, it does so through the investment channel. If it mainly finances consumption, it can even discourage growth (reducing incentive for supplying labor).

1Micro-level studies find strong positive effect of remittances for financing start-up capital in developing countries [11].

2Their estimation shows that remittances have a positive impact on growth but only in countries with less developed financial systems, this effect being zero or even negative for countries with more developed financial systems.

3Acosta et al. [6] use pooled growth regressions for developing countries and assess the growth effects of remittances in Latin America by using a dummy variable approach. The differences in our results could be due to the fact that in our growth regressions, we control for both investment and human capital while they do not. Hence, they attempt to measure the direct effect of remittances on growth while we look at the impact of remittances on investment, and then growth after controlling for investment and other variables of interest.

4For dynamic panels, Anderson and Hsiao [12] propose differencing to remove country specific effects and then using lagged values of the regressors as instruments to generate consistent estimates. Specifically, under the assumption that explanatory variables are predetermined, their lagged values (one-period) are valid instruments and if explanatory variables are endogenous, then two period lagged values are valid.

5In FE, the presence of country specific effect leads to a correlation between a lagged regressor and the error term, generating biased estimates.

6Besides, if explanatory variables are strictly exogenous, then there should be no significant differences between the FE and the GMM results.

References

Select your language of interest to view the total content in your interested language
Post your comment

Share This Article

Relevant Topics

Article Usage

  • Total views: 11674
  • [From(publication date):
    December-2014 - Aug 19, 2017]
  • Breakdown by view type
  • HTML page views : 7868
  • PDF downloads :3806
 

Post your comment

captcha   Reload  Can't read the image? click here to refresh

Peer Reviewed Journals
 
Make the best use of Scientific Research and information from our 700 + peer reviewed, Open Access Journals
International Conferences 2017-18
 
Meet Inspiring Speakers and Experts at our 3000+ Global Annual Meetings

Contact Us

 
© 2008-2017 OMICS International - Open Access Publisher. Best viewed in Mozilla Firefox | Google Chrome | Above IE 7.0 version
adwords