2016 10 * : CHIP2002 2007 2008 2013 OLG : 12% ~13% 16% ~17% 12% ~ 20% 5% ~6% Logit : 2015 2.77 1.69 1 ( 2011; 2014; 2015) 16%~20%(Chenetal2015) ( 2005 2014;Xu & Xie2015; 2015) 2003 28 1 (2004-2020 ) (2015-2040)( ) 2020 1800 2500 2015 2170.5 2415.27 2014 822.6 981.65 1294.73 2015 80% 124.6 2014 * :100872 :huxia@ruc.edu.cn (16CJL014) (CHIP) 25
1 84% 40 73% 52 100%(2015 ) (2015 ) (2012 ) 20 60% : :1 ;2 3 ;3 ;4 ;5 3 ;6 ;7 20 3 3 3 3 3 3 3 ; 3 3 6 6 3 3 3 3 : 2016 8 :(1) 2075.42 2014 1.6 ( 2013; 2015) ;(2) ( 2014;Wu & Zhang 2015) 2015 11 28 ( 2006; 2013) ( 2015 67 ) ( 2016 2015) 26 2016 7 27 2014
( 2016 44 ) ( ) ( 2015) (Meng2003;Giles & Yoo2007; 2008; 2009; 2010; Rosenzweig & Zhang2014; 2014) (Yuan & Zhang2015) (Chenetal2015) ( 2014) ( 2001) ( 2004) (Modigliani& Cao2004;Chamon & Prasad2010) ( 2013; 2014) (Zhao1999) ( 2014) ( 2014) : 1: (CHIP)2002 2007 2008 2013 2016 10 : ( 2011) ; (Pro- pensityscorematchingpsm) (Xu & Xie2015) (Bias) 2 : 2: 27
(5) (6) : (3) : C 2 S(1+r)+P+w kh k (3) h p h k w p 2 w k S 1 r q ( ) h k w k Pi- S / g g =- Us U s / S =- g (1+r)(1+δ) <0 (7) racha & Zhu(2012) (7) g OLG 1 U=lnC 1+δlnC 2 (1) 2 U C 1 C 2 1 2 δ 1 1 (2) : 2 (5) : C 1 w p h p -S-q (2) S / h 2 =- Us k h k U s / S =- w k (1+r)(1+δ) <0 1 P (8) 2 (8) ( ) h k g (Chi- q (4) : nesehouseholdincomeprojectchip)2002 2007 h k=h(gq) (4) 2008 2013 CHIP2002 hk g >0 28 ;CHIP2007 2008 g h k (theinstituteforthe (2) StudyofLaborIZA) ;CHIP2013 (3)(4) ( ) 14 128 ( ) (1) : U U s= S =0 -δ (1+r)C 1+C 2=0 (5) (5) g : U s h k g = C2 g =wk g >0 (6) CHIP 1 CHIP2007 2008 9 15 ( ) ( ) ( ) 1 28
15 18 ( ( ) 3 2010) 5 18 Logit ( ) 2 3. ln( ) ln( ) / 1. ln ( ) ( 1) 20 4 / / 7 2. 46 4. (2014) 20% 2 = / ln( ) ln( ) / / =1 =0 =3 =2 =1 =1 =0 ln( ) / : CHIP2002 2007 2008 2013 2016 10 =1 =0 \ =0 =1 =2 \ \ \ =3 =4 29
; 10. 5.ln( ) U ( ) 3 ln( ) 6. 6 1 2002 ( 2012) -2013 7. Logit 95% 8. 1 9. 3 3 13208.50 8446.07 667 10431.41 6907.58 884 11167.46 7105.86 343 18404.05 18093.24 667 15186.18 14352.75 884 16765.71 18189.27 343 0.82 0.32 667 0.75 0.30 884 0.76 0.38 343 / 0.04 0.11 667 0.13 0.25 884 0.11 0.22 343 1.41 0.62 667 1.65 0.69 884 1.53 0.67 343 2002 0.96 0.21 667 0.85 0.36 884 0.93 0.25 343 3.44 0.67 667 2.39 0.78 884 2.50 0.75 343 1.83 0.46 667 1.77 0.57 884 1.81 0.54 343 0.69 0.47 666 0.64 0.48 884 0.66 0.47 343 9.04 5.49 667 7.09 5.15 884 8.44 5.79 343 1.80 0.89 667 1.73 0.85 884 1.71 0.85 343 36.62 5.39 666 35.81 9.32 884 35.91 8.80 343 30
2016 10 3 27435.60 15309.13 1714 17593.94 12208.38 2535 21800.78 13310.73 1263 39732.33 26865.72 1714 30746.71 21528.76 2535 37182.44 25150.11 1263 0.75 0.26 1714 0.59 0.23 2535 0.61 0.22 1263 / 0.06 0.10 1714 0.17 0.22 2535 0.13 0.18 1263 2007 2008 1.94 0.63 1714 2.03 0.68 2535 2.01 0.68 1263 0.95 0.22 1714 0.49 0.50 2535 0.83 0.38 1263 3.41 0.74 1714 1.65 0.78 2535 2.06 0.74 1263 1.82 0.57 1714 1.49 0.63 2535 1.80 0.61 1263 0.56 0.50 1714 0.22 0.42 2535 0.32 0.47 1263 7.19 5.43 1714 6.52 5.43 2535 6.80 5.34 1263 1.89 0.77 1714 1.93 0.75 2535 1.90 0.75 1263 35.04 7.03 1714 34.70 6.98 2535 35.12 7.08 1263 38043.26 18547.69 284 34749.67 20024.53 418 66954.98 36360.11 284 61936.07 35047.90 418 0.62 0.23 284 0.60 0.23 418 1.39 0.67 284 1.61 0.79 418 0.98 0.14 284 0.69 0.46 418 2013 3.87 0.93 284 2.56 1.26 418 1.88 0.70 284 1.88 0.91 418 0.27 0.44 284 0.21 0.41 418 0.17 0.38 284 0.19 0.40 418 8.06 7.12 284 6.62 8.38 418 2.19 0.86 284 2.46 0.95 418 39.66 6.92 284 38.72 12.09 418 : CHIP2002 2007 2008 2013 2013 418 2013 ln( ) 11 Logit (2014) CHIP2007 Migkid=α+β Citylevel+χ Migspouse +γx+ε (9) Migkid Citylevel 1 Migspouse X 4 Logit 1% U 2002 2003 31
4 :Logit (2014) 2002 1% 2002 (2) (3) (0.097) 1% -0.010 *** 1550 4230 702 ; Pseudo-R 2 0.155 0.240 0.227 log-likelihood -894.882-2169.153-366.398 :* ** *** 10% 5% 1% ( ) Becker 2007 2008 1% OLS : Con=α+β Migkid+γX+ε (10) Con ln( ) ln( ) / Migkid U β β 32 (1) (2) (3) 2002 2007 2008 2013-1.105 *** (0.198) -0.683 *** (0.124) 1.187 *** (0.249) 0.049*** (0.011) -0.081 (0.125) 0.115 (0.341) ln( ) 0.232 ** 0.830 *** (0.138) (0.002) 0.044 (0.071) 0.040 (0.137) -0.416 *** (0.115) -0.005 (0.092) 2.890 *** (0.146) 0.012 * (0.007) 1.000 *** (0.084) 0.306 *** (0.102) -0.036 (0.083) 0.079 ** (0.039) -0.001 * (0.001) (0.052) -0.214 ** (0.084) -0.884 *** (0.262) -0.313 (0.229) 2.807 *** (0.439) 0.042 *** (0.015) -0.085 (0.221) 0.235 (0.245) 0.077 (0.176) 0.609 *** (0.153) -0.008 *** (0.002) -0.224 ** (0.109) -0.439 *** (0.132)
(Giles & Yoo2007;Chenetal2015;Chuetal 2015) X ε ( ) : 5 CHIP2002 CHIP2007 2008 OLS ln( ) ln( ) ln( ) 2016 10 U Rosenzweig & Zhang(2014) (2014) 1% ( ) : (4) (8) / 1% 10% 1 5 : 2002 2007 2008 (1) (2) (3) (4) (5) (6) (7) (8) ln( ) ln( ) / ln( ) ln( ) / 0.116 *** (0.020) 0.184 *** (0.026) 0.142 *** (0.023) -0.100 *** (0.010) 0.170 *** (0.007) 0.525 *** (0.016) 0.293 *** (0.013) -0.107 *** (0.005) ln( ) -0.347 *** (0.025) 0.446 *** (0.027) 0.580 *** (0.030) -0.077 *** (0.023) -0.108 *** (0.013) 0.476 *** (0.058) 0.778 *** (0.015) 0.007 (0.005) 0.014 (0.013) 0.042 *** (0.015) 0.012 (0.014) 0.015 * (0.008) 0.005 (0.005) 0.041 *** (0.012) 0.016 * (0.009) 0.002 (0.003) 0.011 (0.008) 0.047 *** (0.013) 0.015 (0.009) 0.022 *** (0.005) -0.008 * (0.004) -0.031 *** (0.010) -0.009 (0.006) 0.005 ** (0.002) -0.001 *** * *** 0.000 * 0.000 *** 0.000 ** 1550 1550 1550 1550 4233 4233 4233 4233 Adj-R 2 0.275 0.332 0.471 0.113 0.163 0.317 0.509 0.080 log-likelihood -655.011-1050.056-862.463 324.102 135.766-3621.694-2421.322 1234.992 :(1)* ** *** 10% 5% 1% (2) CHIP2007-2008 CHIP2013 OLS PSM CHIP2013 33
9.2% 5 17% = E [Y 1i-Y 0i T i=1] 13.3% ATT 18.8% 26.5% + { E[Y 0i T i=1]-e[y 0i T i=0]} ( 11) T=1 T=0 Y 2 ( ) (PSM) (Rosenbaum & Rubin1983) : 1 2 ( ) ( ) OLS ATT: E[Y i T i=1]-e[y i T i=0] 6 : : (1) (2) (3) (4) :ln( ) (5) (6) (7) (8) 0.092 *** (0.021) 0.133 *** (0.016) 0.188 *** (0.014) 0.265 *** (0.024) 0.175 *** (0.036) 0.239 *** (0.028) 0.331 *** (0.025) 0.403 *** (0.035) ln( ) -0.052 *** (0.018) -0.125 *** (0.018) -0.145 *** (0.013) -0.147 *** (0.030) 0.862 *** (0.042) 0.774 *** (0.040) 0.763 *** (0.024) 0.804 *** (0.047) -0.013 (0.017) 0.014 (0.010) 0.012 (0.009) -0.002 (0.015) -0.025 (0.025) 0.010 (0.020) 0.044 *** (0.014) (0.025) 0.000 (0.011) -0.006 (0.007) 0.005 (0.008) -0.029 *** (0.009) -0.004 (0.019) 0.002 (0.017) 0.004 (0.012) -0.040 *** (0.014) 0.000 0.000 *** 0.000 0.001 *** 458 926 1359 411 457 926 1359 411 Adj-R 2 0.054 0.120 0.168 0.266 0.620 0.419 0.520 0.522 log-likelihood 67.947 78.072-92.036 19.536-191.643-550.283-821.840-176.700 :(1)* ** *** 10% 5% 1% ;(2) CHIP2007 2008 ;(3) 2007 34
Logit 1% ATT OLS Logit 7 12%~ 1% 12%~20% 5%~6% Logit 1 2 2002 2007 2008 3 ( )ATT 7 (11) ATT 13% 16% ~17% :* ** *** 10% 5% 1% 7 Stata sensat 7 2016 10 ATT 2002 2007 2008 ATT T ATT T 0.12 4.02 *** 0.13 12.88 *** ln( ) 0.17 3.93 *** 0.16 6.23 *** ln( ) 0.12 2.81 *** 0.20 8.09 *** / -0.06-4.11 *** -0.05-7.16 *** :CHIP2002 1 2002 :CHIP2007 2008 2 2007-2008 35
(Dehejia & Wah- 1% ba1999) ln( ) ln( ln( ) ) / 1 8 ATT 1% ln( ) ( ) (PSM) 9 8 2002 2007 2008 ATT T ATT T 0.067 4.530 *** 0.166 21.289 *** ln( ) 0.081 4.918 *** 0.162 19.513 *** 0.058 3.775 *** 0.137 15.469 *** 0.058 3.739 *** 0.134 15.004 *** 4 0.065 3.260 *** 0.133 11.693 *** 0.244 8.770 *** 0.317 17.505 *** ln( ) 0.142 4.532 *** 0.282 14.348 *** ln( ) 0.196 6.795 *** 0.27 12.647 *** 0.200 6.762 *** 0.282 13.087 *** 4 0.086 2.34 ** 0.229 8.554 *** :* ** *** 10% 5% 1% 9 : (PSM) 2002 2007 2008 (1) (2) (3) (4) (5) (6) (7) (8) ln( ) ln( ) / ln( ) ln( ) / 0.158 ** (0.079) 0.164 *** (0.034) 0.129 *** (0.030) -0.087 *** (0.015) 0.146 *** (0.010) 0.265 *** (0.021) 0.235 *** (0.017) -0.067 *** (0.007) 0.188 ** (0.084) 0.254 *** (0.063) 0.123 ** (0.053) 0.000 (0.017) -0.011 (0.014) 0.098 *** (0.034) -0.002 (0.026) 0.002 (0.011) * -0.096 (0.128) -0.116 (0.088) -0.060 (0.076) 0.036 (0.023) -0.012 (0.020) 0.020 (0.042) -0.013 (0.035) 0.017 (0.014) ln( ) -0.898 *** (0.175) 0.390 *** (0.032) 0.550 *** (0.036) -0.061 ** (0.028) -0.131 *** (0.018) 0.480 *** (0.038) 0.751 *** (0.017) 0.015 *** (0.004) 0.089 ** (0.041) 0.039 ** (0.017) -0.001 (0.016) 0.013 * (0.008) 0.002 (0.006) 0.039 *** (0.012) 0.009 (0.009) 0.002 (0.003) 0.049 * (0.028) 0.023 (0.015) 0.011 (0.012) 0.015 *** (0.005) -0.005 (0.005) -0.029 *** (0.010) -0.004 (0.007) 0.002 (0.002) -0.001 * *** 0.000 0.000 *** 0.000 1024 1024 1024 1024 2966 2966 2966 2966 Adj-R 2 343 343 343 343 1252 1252 1252 1252 log-likelihood 0.259 0.311 0.467 0.115 0.170 0.315 0.550 0.053 :* ** *** 10% 5% 1% 36
2.77 18 16 4 :(1) CHIP2002 2007 2008 2013 ( OLG ) (2) (3) (4) (5) : 6 12% ~13% htp://edu.bjhd.gov.cn/xw/jwdt/ 16%~17% 12% ~20% 201604/t20160422 1262046.html 5%~6% 5 (2010) ( 2016) 6 Stata 2% Logit 7 Kernel Radius : 2012: 8 ( 2013: ) 1 2010: 1 2016: : No.4(2016) 2013: 28 1860 ( ) 1 2014: : (Balancing Assumption) 3 2016 10 18 / 18 (6) ( ) (7) (8) winsorize 6 : 2014: U 1 2015 A01 htp://www.stats.gov.cn/2016 4 28 2014: : 2 (CommonSupportAssumption) 3 2013: ( ) 3 18 2015: 37
CHIP2008 ChuR.M.Liu & G.J.Shi(2015) Howrural-urbaniden- 5 2015: Chinesemigrantworkers Asia PacificJournalof Mar- 4 2014: ciation94(448):1053-1062. 1 2014: networksandhouseholdconsumptiondecisions:anempirical 10 2001: view of Economicsand Statistics89(3):534-551. 11 2010: : andprecautionarysavinginurbanchina Journalof Com- 1 2011: ModiglianiF.& S.L.Cao(2004) The Chinesesaving 1 2009: PirachaM.& Y.Zhu(2012) Precautionarysavingsbynatives 8 2006: 2767-2776. 7 2015: tivitytoanunobservedbinarycovariateinanobservational 2008 CHIP 5 2008: RosenzweigM.& J.Zhang(2014) Co-residencelife-cycle savingsand inter-generational supportin urban China 2 2014: WuX. & Z.Zhang(2015) Population migration and 4 2015: : PopulationStudiesCenterResearchReportNo.13-794. 3 2014: migrationonchildren s wel-beingin China European CHIP SociologicalReview31(4):502-519. ( ) 1 ChamonM.D.& E.S.Prasad(2010) Whyaresavingratesof urban householdsin China rising American Economic Journal:Macroeconomics2(1):93-130. ChenB.M.Lu & N.Zhong(2015) Howurbansegrega- tion distorts Chinese migrants consumption World Development70:133-146. tificationinfluencesconsumptionpaterns Evidencefrom keting and Logistics27(1):40-60. 2004: DehejiaR.H.& S.Wahba(1999) Causalefectsinnon- experimentalstudies:reevaluatingtheevaluationoftrain- GilesJ.& K.Yoo(2007) Precautionarybehaviormigrant ingprograms Journalofthe American Statistical Aso- analysisusinghouseholdpaneldatafromruralchina Re- MengX.(2003) Unemploymentconsumptionsmoothing parative Economics31(3):465-485. puzzleandthelife-cyclehypothesis Journalof Economic Literature42(1):145-170. andimmigrantsin Germany Applied Economics44(21): RosenbaumP.R.& D.B.Rubin(1983) Assessingsensi- studywithbinaryoutcome Journalofthe RoyalStatis- ticalsocietyseriesb.45(2):212-218. NBER WorkingPaperNo.20057. children sschoolenrolmentsin China1990-2005 XuH.& Y.Xie(2015) Thecausalefectsofrural-to-urban YuanC.& L.Zhang(2015) Publiceducationspending andprivatesubstitutioninurbanchina Journalof De- velopmenteconomics115(c):124-139. ZhaoY.(1999) Leavingthecountryside:Rural-to-urban migrationdecisionsinchina American Economic Review 89(5):281-286. ( : ) 38