Does provincial economic development in vietnam foster internal migration

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS DOES PROVINCIAL ECONOMIC DEVELOPMENT IN VIETNAM FOSTER INTERNAL MIGRATION? o0o -Thesis Instructor: Dr Pham Khanh Nam Student: Võ Văn Hưng CLASS 20 Ho Chi Minh City, September 2015 CERTIFICATION “I certify the content of this dissertation has not already been submitted for any degree and is not being currently submitted for any other degrees I certify that, to the best of my knowledge, any help received in preparing this dissertation and all source used, have been acknowledged in this dissertation.” Signature Vo Van Hung Date: 14th September 2015 ACKNOWLEDGEMENT I would like to express my deepest gratitude to my instructors in Vietnam Netherlands Program, who helped me to achieve the knowledge through interesting lessons, useful assignment, utility seminars and new information during my master studying I greatly express my special thanks to Dr Pham Khanh Nam for all his academic recommendations through finishing thesis process I am grateful to all the staffs in the program have helped me to reach books and necessary documents during the learning process My thanks are also extended to all my classmates, who have companions and share learning experiences with me through over last years Mental support from my family is one of the most bolster for my effort to finish this program I would like to send my thanks and my love to all my family’s members ABBREVIATIONS PGDP Provincial Gross Domestic Product PCI Provincial Competiveness Index GSO General Statistics Office of Vietnam SEM Structural Equation Modeling FEM Fixed Effect Model REM Random Effect Model OLS Ordinary Least Squares VIF Variance Inflation Factor HCMC Ho Chi Minh City VND Vietnam Dong (Vietnamese Currency) ABSTRACT This paper investigates the impact of provincial economic development on internal migration in Vietnam based on the complex casual relationship between migration and economic development We examine that impact for further understanding about the determinants could lead to human migration decision beyond the difficulties at the new destinations Tabulations data about migration flows from GSO database in duration 2005-2013 and national statistical datasets from Statistical Yearbook about provincial development are used to analyze the impact of economic development on migration The quantitative results of Structural Equation Modeling (SEM) and Common Panel Methods indicated that the higher economic development provinces have higher immigration attractiveness Key Words: Migration, economic development, Vietnam, population, origin areas, destination areas, employment, jobs TABLE OF CONTENTS CHAPTER 1: INTRODUCTION 1.1 Problem statement 1.2 Research objectives and research questions 1.3 Scope of the study 1.4 Organization of the thesis CHAPTER 2: LITERATURE REVIEW 2.1 Concepts of migration 2.2 Migration theories 2.2.1 Harris – Todaro theory 2.2.2 Lee’s theory 2.2.3 Lewis’s theory 2.2.4 Other cognizance about migration 10 2.3 Empirical studies on migration 11 2.3.1 Economic development and other factors 11 2.3.2 Determinants of migration 13 2.3.3 Linkages between economic development and migration 14 CHAPTER 3: RESEARCH METHODOLOGY 3.1 Analytical framework and hypotheses 17 3.2 Estimation methods 21 3.2.1 Structural equation model (SEM) 21 3.2.2 Other common panel data methods 23 a Pooled OLS model 23 b Fixed effects model (FEM) 24 c Random effects model (REM) 24 3.2.3 Testing for appropriate models 25 a F-test for choosing FEM or OLS 25 b Breusch - Pagan LM-test for choosing REM or OLS 25 c Hausman test for choosing FEM or REM 26 3.3 Model specification 26 3.3.1 Structural equation model (SEM) 27 3.3.2 Common panel data techniques 29 3.3.3 Migration criteria 30 3.3.4 Instrument variable 31 3.4 Data sources 32 CHAPTER 4: EMPIRICAL RESULTS 4.1 Overview about inter-provincial migration in Vietnam 34 4.2 Descriptive statistics 37 4.3 Regression results 41 4.3.1 Results from simultaneous equation method (SEM) 41 4.3.2 Robustness test - common panel data methods 44 a Ordinary Least Squared Method (OLS) 44 b Fixed Effect Model (FEM) .47 c Random Effects Model (REM) 49 CHAPTER 5: CONCLUSION AND POLICY IMPLICATIONS 5.1 Main findings 53 5.2 Limitation of the study 54 5.3 Policy implications 54 5.4 Suggestions for further studies 55 References 56 APPENDIX 60 CHAPTER INTRODUCTION 1.1 Problem statement The modernization of the country is always an important work in the context of the economy increasingly dynamic and society moves deeper in integration with the world One of the top priorities for a sustainable development and modern society is to reduce social inequality and eliminate poverty Along with the existence and development of the country, leaders, policy makers and scientists always seek for better solutions and improve the policies in order to match those requirements However, in a most natural way, it has never been a simple task and can be done in the short - term to get the immediate results but it is a difficult and long - term process Vietnam never could be an exception case in suffering the negative effect from development process These negative effects are the huge factors causing inequality, poverty, stagnation, unkempt for the social and economic development not only in the present but also for future generations One downside is that people often together rushed to the area where the development which is generally higher than other areas, as an allusion proverb “lua thoc dau bo cau theo do” in Vietnamese It is easily to espy the phenomenon of rural people leave their fields for the big cities to find jobs The nature of that phenomenon can be interpreted as the migrant attempt to come to higher advantage destinations with the expectation of improving their lives (De Jong & Fawcett, 1981) However, the out-control migration cause the imbalance in population density such as overload status at the large city or the slow growth rate in rural areas since the lack of labor In addition, Migration leads to many serious social problems For instance, mass-migration could create environmental pollution because of the inadequate infrastructure (Bilsborrow, 1992) and theft could also be a painful problem for downtown citizen (Deshingkar, 2006) because the migrants fall into deadlock situation There are enormous difficulties for any migrant when they have to resettle with many different conditions at the new destinations, but in reality a huge amount of APPENDIX The correlation matrix among variables corre urban income in_struct density pgdp edu health pop (obs=482) urban income in_str~t density pgdp edu 1.0000 health urban income 1.0000 0.4205 1.0000 in_struct density -0.4298 0.4137 -0.1397 0.2610 1.0000 0.2043 1.0000 pgdp edu 0.5183 -0.3692 0.5820 -0.2540 -0.1249 0.0267 0.7336 -0.4707 1.0000 -0.3377 health -0.0098 -0.0945 0.0565 -0.1168 -0.0509 0.6603 1.0000 pop 0.3698 0.1992 -0.0104 0.7525 0.7974 -0.4257 -0.1738 pop 1.0000 SEM regression result Structural equation model Number of obs Estimation method = ml Log likelihood = -25894.334 = 463 OIM Coef Std Err z P>|z| [95% Conf Interval] Structural urban F R-squared Adj R-squared Root MSE P>|t| 0.033 0.001 0.000 0.043 0.095 0.000 0.251 = = = = = = 497 132.20 0.0000 0.6181 0.6135 15901 [95% Conf Interval] -175.7601 2.837012 0804412 -23258.46 -426.9203 4.554085 -12678.37 -7.263587 11.77006 1815947 -351.2734 5280.875 9.86386 3321.701 Result from model by pool OLS method reg in_mig in_struct density pgdp edu health pop Source SS df MS Model Residual 2.0912e+11 1.1501e+11 475 3.4853e+10 242121632 Total 3.2412e+11 481 673853088 in_mig Coef in_struct density pgdp edu health pop _cons -75.99597 15.1619 0510512 -6522.622 2950.735 8.514016 -8883.129 Std Err 12.65141 2.485184 0183724 5695.853 1451.862 1.124378 3329.149 t -6.01 6.10 2.78 -1.15 2.03 7.57 -2.67 Number of obs F( 6, 475) Prob > F R-squared Adj R-squared Root MSE P>|t| 0.000 0.000 0.006 0.253 0.043 0.000 0.008 = = = = = = 482 143.95 0.0000 0.6452 0.6407 15560 [95% Conf Interval] -100.8556 10.27859 0149499 -17714.81 97.86746 6.304645 -15424.81 -51.13632 20.04522 0871524 4669.563 5803.602 10.72339 -2341.449 Result from model by pool OLS method reg in_mig urban income in_struct density pgdp edu health pop Source SS df MS Model Residual 2.1506e+11 1.0906e+11 473 2.6883e+10 230571861 Total 3.2412e+11 481 673853088 Std Err t Number of obs F( 8, 473) Prob > F R-squared Adj R-squared Root MSE P>|t| = = = = = = 482 116.59 0.0000 0.6635 0.6578 15185 in_mig Coef [95% Conf Interval] urban income in_struct density 313.8702 -97.24901 -36.21534 11.16243 64.35439 42.79937 14.65264 2.567677 4.88 -2.27 -2.47 4.35 0.000 0.024 0.014 0.000 187.4143 -181.3494 -65.00766 6.116968 440.326 -13.14858 -7.42302 16.2079 pgdp edu health pop _cons 075221 940.3558 622.6858 7.841474 -13125.93 02611 6006.728 1490.666 1.294849 4206.638 2.88 0.16 0.42 6.06 -3.12 0.004 0.876 0.676 0.000 0.002 023915 -10862.82 -2306.461 5.297106 -21391.94 126527 12743.53 3551.833 10.38584 -4859.919 Result from model by FEM xtreg in_mig urban density pgdp edu health pop, fe Fixed-effects (within) regression Group variable: province2 Number of obs Number of groups = = 497 63 R-sq: Obs per group: = avg = max = 7.9 within = 0.2947 between = 0.6065 overall = 0.5577 corr(u_i, Xb) F(6,428) Prob > F = -0.9770 in_mig Coef urban density pgdp edu health pop _cons 190.2177 31.68295 -.2475056 -4630.607 6113.935 78.4641 -120273.1 122.9007 15.34816 0212722 9030.046 1266.075 9.564696 11105.56 sigma_u sigma_e rho 73537.88 7226.8822 99043455 (fraction of variance due to u_i) F test that all u_i=0: Std Err F(62, 428) = t 1.55 2.06 -11.64 -0.51 4.83 8.20 -10.83 28.36 P>|t| = = 0.122 0.040 0.000 0.608 0.000 0.000 0.000 29.81 0.0000 [95% Conf Interval] -51.34643 1.515814 -.2893166 -22379.36 3625.437 59.66448 -142101.3 431.7818 61.85009 -.2056945 13118.15 8602.433 97.26372 -98444.87 Prob > F = 0.0000 Result from model by FEM xtreg in_mig income density pgdp edu health pop, fe Fixed-effects (within) regression Group variable: province2 Number of obs Number of groups = = 497 63 R-sq: Obs per group: = avg = max = 7.9 within = 0.4256 between = 0.5986 overall = 0.5524 corr(u_i, Xb) F(6,428) Prob > F = -0.9866 in_mig Coef income density pgdp edu health pop _cons 389.5499 53.14346 -.405755 -10928.11 2248.123 98.35146 -142612.2 38.86488 13.9593 0245181 8159.962 1170.351 8.762794 10257.31 sigma_u sigma_e rho 98337.651 6521.9433 99562066 (fraction of variance due to u_i) F test that all u_i=0: Std Err F(62, 428) = t 10.02 3.81 -16.55 -1.34 1.92 11.22 -13.90 40.08 P>|t| = = 0.000 0.000 0.000 0.181 0.055 0.000 0.000 52.85 0.0000 [95% Conf Interval] 313.1601 25.70615 -.4539459 -26966.7 -52.228 81.128 -162773.2 465.9397 80.58076 -.357564 5110.475 4548.474 115.5749 -122451.2 Prob > F = 0.0000 10 Result from model by FEM xtreg in_mig in_struct density pgdp edu health pop, fe Fixed-effects (within) regression Group variable: province2 Number of obs Number of groups = = 482 63 R-sq: Obs per group: = avg = max = 7.7 within = 0.3197 between = 0.5962 overall = 0.5550 corr(u_i, Xb) F(6,413) Prob > F = -0.9808 in_mig Coef in_struct density pgdp edu health pop _cons -56.1361 34.26454 -.2785607 -2357.575 7045.905 87.20347 -127663.6 34.18615 15.41284 0221343 9024.853 1206.996 9.612042 11434.29 sigma_u sigma_e rho 81511.167 7197.5538 99226318 (fraction of variance due to u_i) F test that all u_i=0: Std Err F(62, 413) = t -1.64 2.22 -12.59 -0.26 5.84 9.07 -11.16 29.15 P>|t| = = 0.101 0.027 0.000 0.794 0.000 0.000 0.000 32.35 0.0000 [95% Conf Interval] -123.3366 3.967136 -.3220706 -20097.95 4673.283 68.30885 -150140.2 11.06445 64.56194 -.2350507 15382.8 9418.527 106.0981 -105186.9 Prob > F = 0.0000 11 Result from model by FEM xtreg in_mig urban income in_struct density pgdp edu health pop, fe Fixed-effects (within) regression Group variable: province2 Number of obs Number of groups = = 482 63 R-sq: Obs per group: = avg = max = 7.7 within = 0.4326 between = 0.5900 overall = 0.5501 corr(u_i, Xb) F(8,411) Prob > F = -0.9873 in_mig Coef urban income in_struct density pgdp edu health pop _cons -257.1457 417.0454 -47.93839 56.85476 -.4245064 -8920.383 2797.499 102.55 -142766.9 121.785 46.50165 31.60603 14.40836 0264106 8316.409 1242.287 9.068355 10610.91 sigma_u sigma_e rho 102001.75 6589.5776 99584385 (fraction of variance due to u_i) F test that all u_i=0: Std Err F(62, 411) = t -2.11 8.97 -1.52 3.95 -16.07 -1.07 2.25 11.31 -13.45 33.88 P>|t| = = 0.035 0.000 0.130 0.000 0.000 0.284 0.025 0.000 0.000 39.16 0.0000 [95% Conf Interval] -496.5449 325.6347 -110.068 28.5315 -.4764231 -25268.39 355.4707 84.72385 -163625.3 -17.74645 508.4562 14.19124 85.17802 -.3725898 7427.621 5239.527 120.3761 -121908.5 Prob > F = 0.0000 12 Result from model by REM xtreg in_mig urban density pgdp edu health pop, re Random-effects GLS regression Group variable: province2 Number of obs Number of groups = = 497 63 R-sq: Obs per group: = avg = max = 7.9 within = 0.1881 between = 0.6653 overall = 0.6139 corr(u_i, X) Wald chi2(6) Prob > chi2 = (assumed) in_mig Coef Std Err z urban density pgdp edu health pop _cons 403.7091 14.27687 -.0762968 1136.121 3290.375 15.94456 -33701.26 91.8202 5.002836 0134389 7678.078 1232.969 2.403505 5836.968 sigma_u sigma_e rho 12661.966 7226.8822 75428352 (fraction of variance due to u_i) 4.40 2.85 -5.68 0.15 2.67 6.63 -5.77 P>|z| 0.000 0.004 0.000 0.882 0.008 0.000 0.000 = = 195.70 0.0000 [95% Conf Interval] 223.7448 4.471496 -.1026366 -13912.63 873.8008 11.23378 -45141.51 583.6734 24.08225 -.049957 16184.88 5706.95 20.65534 -22261.02 13 Result from model by REM xtreg in_mig income density pgdp edu health pop, re Random-effects GLS regression Group variable: province2 Number of obs Number of groups = = 497 63 R-sq: Obs per group: = avg = max = 7.9 within = 0.2567 between = 0.5910 overall = 0.5493 corr(u_i, X) Wald chi2(6) Prob > chi2 = (assumed) in_mig Coef Std Err z income density pgdp edu health pop _cons 187.2755 18.63197 -.1249037 814.3932 2470.674 18.31686 -29025.75 39.61928 4.872776 017484 7604.27 1276.334 2.38371 5585.759 sigma_u sigma_e rho 11151.656 6521.9433 74513498 (fraction of variance due to u_i) 4.73 3.82 -7.14 0.11 1.94 7.68 -5.20 P>|z| 0.000 0.000 0.000 0.915 0.053 0.000 0.000 = = 203.48 0.0000 [95% Conf Interval] 109.6231 9.08151 -.1591718 -14089.7 -30.89505 13.64488 -39973.63 264.9278 28.18244 -.0906357 15718.49 4972.244 22.98885 -18077.86 14 Result from model by REM xtreg in_mig in_struct de nsity pgdp edu health pop, re Random-effects GLS regressio n Group variable: province2 Number of obs Number of groups = = 482 63 R-sq: Obs per group: = avg = max = 7.7 within = 0.2057 between = 0.6400 overall = 0.6024 corr(u_i, X) Wald chi2(6) Prob > chi2 = (assumed) in_mig Coef Std Err z in_struct density pgdp edu health pop _cons -81.24489 22.29229 -.0837568 -849.7024 4860.718 15.74508 -24916.07 24.02343 5.369721 0141084 7855.039 1230.053 2.506244 5866.425 sigma_u sigma_e rho 12830.085 7197.5538 76062392 (fraction of variance due to u_i) -3.38 4.15 -5.94 -0.11 3.95 6.28 -4.25 P>|z| 0.001 0.000 0.000 0.914 0.000 0.000 0.000 = = 178.96 0.0000 [95% Conf Interval] -128.33 11.76784 -.1114087 -16245.3 2449.857 10.83294 -36414.06 -34.15984 32.81675 -.0561048 14545.89 7271.578 20.65723 -13418.09 15 Result from model by REM xtreg in_mig urban income in_struct density pgdp edu health pop, re Random-effects GLS regression Group variable: province2 R-sq: within between overall corr(u_i, X) = 0.2333 = 0.6585 = 0.6214 = (assumed) Std Err z Number of obs Number of groups = = 482 63 Obs per group: avg max Wald chi2(8) Prob > chi2 = = = = = 7.7 220.77 0.0000 in_mig Coef urban income in_struct density pgdp edu health pop _cons 227.8549 157.4284 -55.02182 19.84327 -.1225855 4246.737 2208.191 16.88592 -30846.25 102.9276 44.17028 25.00604 5.279713 0179725 7665.721 1316.07 2.392575 5964.156 sigma_u sigma_e rho 11137.02 6589.5776 7406925 (fraction of variance due to u_i) 2.21 3.56 -2.20 3.76 -6.82 0.55 1.68 7.06 -5.17 P>|z| 0.027 0.000 0.028 0.000 0.000 0.580 0.093 0.000 0.000 [95% Conf Interval] 26.12043 70.85623 -104.0328 9.495226 -.1578109 -10777.8 -371.2589 12.19656 -42535.78 429.5893 244.0006 -6.010888 30.19132 -.08736 19271.27 4787.641 21.57528 -19156.72 16 Breusch – Pagan LM test for choosing appropriate model between REM and OLS Breusch and Pagan Lagrangian multiplier test for random effects in_mig[province2,t] = Xb + u[province2] + e[province2,t] Estimated results: Var in_mig e u Test: Var(u) = sd = sqrt(Var) 6.74e+08 4.34e+07 1.24e+08 chibar2(01) = Prob > chibar2 = 25958.68 6589.578 11137.02 677.48 0.0000 17 Hausman test for choosing appropriate model between REM and FEM Coefficients (b) (B) fixed random urban income in_struct density pgdp edu health pop -257.1457 417.0454 -47.93839 56.85476 -.4245064 -8920.383 2797.499 102.55 227.8549 157.4284 -55.02182 19.84327 -.1225855 4246.737 2208.191 16.88592 (b-B) Difference sqrt(diag(V_b-V_B)) S.E -485.0005 259.617 7.083427 37.01149 -.301921 -13167.12 589.308 85.66408 65.09603 14.53925 19.32974 13.40617 0193522 3224.807 8.747036 b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic chi2(7) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 187.18 Prob>chi2 = 0.0000 (V_b-V_B is not positive definite) ... of provincial economic development on internal migration to interpret whether high economic development provinces attract migrants and whether the encouragement of economic development – an instrument... spontaneous migrants In terms of time, inter -provincial migration in Vietnam has three types, including long-term migration, migration in short-term (temporary migration) and seasonal migration The... stands for provincial competitiveness index, is used to assess the ease of economic governance in incentive investment or doing business of 63 provinces and cities in Vietnam issued by Vietnam Chamber