17 3 2018 4 ( ) ChinaEconomicQuarterly Vol.17,No.3 April,2018 1*, 2005 1%,,,,,,, DOI:10.13821/j.cnki.ceq.2018.04.02,, 20 70,,,,, (,2008;,2013),,, (,2011;,2010),, :, * 1, ;, :, 305,98195; :1-1206-48706429;E-mail:zogcee@gmail.com, cczhuang@uw.edu (71573003) (16JDLJB001)
898 ( ) 17 ;,, (quantity-qualitytrade-of) Gary Becker BeckerandLewis(1973) [Wilis(1973) ], (budgetconstraint) (creditmarketfailure),,, ( ),,, (Glicketal., 2005;Lietal.,2008;RosenzweigandZhang,2009;Hatonand Martin, 2010;Liu,2014), (Blacket al.,2005;park and Chung,2012) (Qian,2009; LordanandFrijters,2013), Blacketal.(2005),,, ;,, ( ) ParkandChung(2012) MHSS,, ;,,, (, ) Qian (2009) CHNS ;, LordanandFrijters(2013), ( ),,,,
3 : 899 ;,,, (IV), IV (Glicketal.,2005;Liet al.,2008;angristetal.,2010;hatonand Martin,2010) (Qian,2009;ParkandChung,2012;Cameronetal.,2013),,,,, IV, IV :,,, ;,,, (regressiondiscontinuitydesign,rdd) RDD,,,,, IV,RDD,,, RDD,, RDD (natural experiment), 2005 1% RDD, 20 80, Q-Q RDD RDD, (DID) :(1) (RDD), ;(2) BeckerandLewis(1973) Wilis (1973), ;(3)
900 ( ) 17, ; (4), : ; ;, ; ;, 1 1 1949 1961 (1953 8 11 ), (1955 3 1 ) (1956 9 16 ) 1962 1978 (1962 12 18 ), (1970 5 20 ) 1975 (1974 12 29 ) (1978 2 24 ) 1979 1991 (1978 3 5,1982 12, 4 ) (1980 9 25 ) (1982 2 9 ) 1992 2000 (1995 2000) (1995 1, 14 ) (2000 3 2 ) 2001 2012 (2001 12 29, ) (2006 12 17 ) 2013, (2013 11 15 ) (2015 10 26 ) ( ) (2015 12 21 ) : (2003) (2010),
3 : 901, :,, :, ;,, RDD ; 1 1, RDD, RDD 20 70, : (Zhang and Spencer,1992;Li,1995;,2003;,2010; Cameronetal.,2013) (,2003)2 2,, :, ( );,, 3 10 (Banister,1987;Cooneyetal.,1991;LiandCooney,1993) 3 3 :, 4 4, 1979 (, 1979 610 ), 1980 (,2014),, 1980 10 1 1 (complier),, 2 2 3 1979 3 1980,, ;,, (Ebenstein,2010) 4 4, ( )
902 ( ) 17,,, 1 2005 1% 1980 10, 1 :, ; (RDD) (DID) RDD,, (treatmentefect) ( ) 0 1, (AngristandLavy, 1999;VanderKlaauw,2002;,2010;,2011; Chenetal.,2013)
3 : 903 (assignmentvariable)x ( 1980 10 ) 1 5, (outcomevariable)y ( ), (treatmentvariable)d ( D=1, D=0) (1) ( ) : τ= lim x c lim x c E[Y X]-lim E[Y X] x c E [ D X ]-lim E [ D X ]. (1) x c Wald c, c c ( ±h),, (1) (complier,,, 26 ) (localaveragetreatmentefect,late) (1), (IV), (ImbensandLemieux,2008;LeeandLemieux,2010) RDD ( T, T=1, T=0), T D IV, (2SLS) D Y 37 : Y =θ+τd +f(x -c)+ε, (2) D =γ+δt +g(x -c)+u. (3) (2) (3) 2SLS,δ ;γ θ ;f( ) g( ), ; X T D, f( ) g( ) ;u ε, T, τ : 5 1, (bin) 6 2, (defier), (monotonicity) ;,,, 7 3, RDD,γ=0,α=0, δ=1
904 ( ) 17 τ= N i=1 N i=1 (T i - 珡 T)(Y i - 珚 Y) (T i - 珡 T)(X i - 珡 X) = E [Y c< X c+h]-e[y c-h X c] E[D c< X c+h]-e[d c-h X c]. (4) (4) (1), ( ), (kerneldensity estimation), (one-sided kernel regression) (pointestimate), Wald : lim x c lim x c lim x c lim x c E[Y X]= E[Y X]= E[D X]= E[D X]= Y i K((X i -x)/h) i:x i >c K((X i -x)/h), ( 5) i:x i >c Y i K((X i -x)/h) i:x i <c K((X i -x)/h), ( 6) i:x i <c D i K((X i -x)/h) i:x i >c K((X i -x)/h), ( 7) i:x i >c D i K((X i -x)/h) i:x i <c i:x i <c K ((X i -x)/h). (8) (5) (8) h ImbensandKalyanaraman(2012) (mean squarederror) (optimalbandwidth) K ( ) (kernelfunction), (trianglekernel) (FanandGijbels,1996;LeeandLemieux,2010),, (rectangularkernel) RDD, (,covariate, W ), LATE (LeeandLemieux,2010), RDD :, RDD,, RDD ;,
3 : 905 (samplingvariability), RDD, 8 1 DID RDD : D, (T=0) (T=1),,, ( ) (averagetreatmentefect,ate) Y =β 0 + β 1D + β 2T + β 3 (D T)+β 4W +e. (9) (9),,(D T), β 4 2005 1%,, 2000, 1%, ( ), 31 ( ) 1300 1/5, 260, ( ) 1980 10 5 ( ) 29, (2003) 8 1 : Y W, Y Y * =Y-Y LeeandLemieux(2010), W D ( W ), W Y τ, W, RDD 9 2 5, 3 ( 1980 10 ), 3
906 ( ) 17 1 10 ;,, ;,, 5,, 30681, 1980 10 3 20187 ( 3, ) 2 3,, 211,, ( 11.96, 13.76 ) (22 28 ),1980 10 12 3,, (compensatedyearsofeducation),,,, (Bucklesand Hungerman,2013; AngristandKrueger,1991),,,,,, 2 1 (bin),, (1980 10 ), 13 4 10 1 (2003),,, 11 2, ( ), ( 14 ), 12 3 2010 2015, 2005, 13 4,,,
3 : 907 2 eduyear yhat ( ) 12.66 11.96 13.76 * (2.889) (2.894) (2.517) ( ) 12.68 11.97 13.81 * (2.883) (2.873) (2.515) nosibs ( =1, =0) 0.387 0.000 1.000 * (0.487) (0.000) (0.000) healthy married male ( =1, =0) 0.997 0.996 0.997 (0.057) (0.060) (0.051) ( =1, =0) 0.384 0.435 0.302 * (0.486) (0.496) (0.459) ( =1, =0) 0.504 0.503 0.506 (0.500) (0.500) (0.500) N 20187 12375 7812 : * 5% ; 2 :, ; RDD, (covariate),
908 ( ) 17 14 1, (GB/T2260), (2010) (1,0 ) 2, ( ) ( ) - (education-healthgradient) (Smith,2004),,, RDD, (Zengetal.,1993;DasGupta,2005; Ebenstein,2010;Lietal.,2011), (3 ), ; ( ),,,, ( 43.5% 30.2%),, (entropybalancing)2 15,, RDD, ( ) 1 2, RDD LeeandLemieux(2010), RDD (1) (5) (8) LATE, 3 (1) ;, RDD, 3 (2) RDD, ;, 3 14 1,, 15 2, ( ), Hainmueler(2012)
3 : 909 (3) (rectangularkernel) RDD 3 (4), 16 1 3 RDD (1) (2) (3) (4) 0.287 *** 0.295 ** 0.383 *** 0.251 ** (0.099) (0.120) (0.101) (0.115) 0.049 *** 0.055 *** 0.0648 *** 0.045 ** (0.017) (0.020) (0.0174) (0.021) LATE 5.888 ** 5.335 ** 5.922 *** 5.551 * (2.396) (2.560) (1.849) (3.031) 1980 10 1980 10 1980 10 1980 10 2.4 2.3 1.9 1.9 30681 30681 30681 30681 : *** ** * 1% 5% 10% ; (standarderror) bootstrap ; LATE( Wald ) ; ImbensandKalyanaraman(2012) ; (4),, 3 (2) (4) 3 (1) :, 5% ( Wald ); 2005 ( ), 0.3 ( Wald ) Wald,, (LATE), ( ) 3,, 5.5,, - BeckerandLewis(1973),,, 1,,, 16 1 1 2 3 :, RDD,,
910 ( ) 17,,,,, :(1),, -,,, ( ) (2),, ;RDD LATE,,, ( ), (3) LATE ( ) (cumulativeefect), (marginal efect), ;, LATE ( ) ( ), RDD,, RDD, ( IV,RDD, ),,,, 1 17 17 1,, ( RDD ), ( RDD ),, RDD
3 : 911,, (McCrary,2008;ImbensandLemieux,2008):,, McCrary(2008) LeeandLemieux(2010), 3 ( ),,, McCrary(2008), 0.012( 0.042), ( ), 3 : McCrary(2008) DCdensity, 95%, ( ) :,, RDD (locallinearregression),, 4,
912 ( ) 17 4 ( RDD ) 1 2 3 male -0.031-0.028-0.019-0.000 (0.023) (0.027) (0.020) (0.016) married 0.039 0.043 0.022 0.001 (0.023) (0.026) (0.019) (0.015) healthy 0.002-0.000-0.001-0.001 (0.003) (0.003) (0.002) (0.002) :, RDD, ; 3 (1), RDD ( ):, RDD, (1980 10 ) 1, 5 RDD,,LATE, 3 RDD, ( ) 5 RDD 1979 10 1980 4 1981 4 1981 10 0.089 0.086-0.030-0.022 (0.092) (0.103) (0.132) (0.113) 0.028 * 0.008 0.036 0.009 (0.015) (0.017) (0.0229) (0.020) LATE 3.136 10.66-0.820-2.492 (3.223) (22.78) (3.865) (15.51) 30681 30681 30681 30681 :(1), 1980 10 6 12 6 12 ;(2) LATE ( ) ( ) 3 (1) : 18 24,, :, 5,, ;,RDD (complier),, RDD, LATE, 2SLS,
3 : 913 (2) (3), 1980 10 (T), 2SLS 6 ( 6 3 (1) (2),, ) 3 (1) (2), 2SLS LATE, 6 RDD (IV/2SLS ) 2.4 (1) 1 (2) 2 (3) 3 (4) 4.729 *** 5.540 * 5.685 *** 4.338 *** (1.777) (3.058) (1.662) (1.008) 6.034 * 4.379 * 7.016 ** 5.965 * (3.239) (2.452) (3.476) (3.170) 16276 7040 14189 20174 : *** ** * 1% 5% 10% ; (robuststandard error); (2) D ( ); (1) (2.4 ) ImbensandKalyanaraman(2012) ( 3 (1) ) ( ) DID ( ) 7 (9) : 6, ;, LATE, (ATE), RDD, ( ),,,, 7 DID 2.4 (1) 2 (2) 3 (3) 5 (4) 0.411 *** 0.327 *** 0.449 *** 0.616 *** (0.086) (0.092) (0.078) (0.065) 0.267 *** 0.189 ** 0.320 *** 0.504 *** (0.082) (0.088) (0.074) (0.063) 16276 14189 20174 30681 : *** ** * 1% 5% 10% ; (robuststandard error); (9) D T ; (1) (2.4 ) ImbensandKalyanaraman(2012) ( 3 (1) )
914 ( ) 17 ( ) (placebo), RDD,, 1999 : (2011), 1980,, :, ;,, 3 (1) :,, (, ), 8,, 8 2.3 (1) 1 (2) 2 (3) 3 (4) 0.014 0.030 0.009-0.044 (0.208) (0.324) (0.227) (0.183) -0.008-0.023-0.013-0.007 (0.023) (0.035) (0.025) (0.020) LATE -1.590-1.319-0.678 6.095 (25.98) (14.63) (18.31) (26.37) 1980 10 1980 10 1980 10 1980 10 10035 2306 4609 6717 :(1) 3 (1) ; (2) *** ** * 1% 5% 10% ; (standarderror) bootstrap ; LATE( Wald ) (triangle kernel) ; (1) ImbensandKalyanaraman(2012), 3 (1) ( ),
3 : 915,, 9,, RDD,,, 18 1 9 2.4 (1) 1 (2) 2 (3) 3 (4) 0.033 0.061 0.022 0.020 (0.066) (0.102) (0.073) (0.060) 0.008 0.013 0.008 0.008 (0.008) (0.012) (0.009) (0.007) LATE 4.434 4.849 2.760 2.624 (9.066) (8.464) (8.914) (7.697) 1980 10 1980 10 1980 10 1980 10 61417 13070 26175 37996 :(1) 3 (1) ;(2) *** ** * 1% 5% 10% ; (standarderror) bootstrap ; LATE( Wald ) (trianglekernel) ; (1) ImbensandKalyanaraman(2012) :,,, -,, ( ), ( ) 10, RDD, 5% 1%; 18 1, 2000 1999,,
916 ( ) 17, ;,RDD 10 RDD 0.393 *** 0.561 *** 0.204 (0.114) (0.172) (0.156) 0.071 *** 0.059 ** 0.082 *** (0.020) (0.029) (0.030) LATE 5.500 *** 9.548 ** 2.504 (1.910) (4.863) (1.921) 16535 8750 7785 :(1), ( );(2) 1980 10, ImbensandKalyanaraman(2012) ;(3) LATE ( ) ( ) 3 (1),,,,,,, ( ), (Zengetal.,1993;Das Gupta,2005;Ebenstein,2010;Lietal., 2011),, 3,,, RDD,,, BeckerandLewis(1973), Q-Q,, ( ), ( ), - 10,,LATE,,
3 : 917 19 1 ( ), 2 20 11, RDD, LATE, Q-Q 11 1979 1981 RDD 0.191 0.346 *** 0.250 * 0.293 ** (0.141) (0.130) (0.138) (0.128) 0.045 * 0.051 ** 0.047 * 0.047 ** (0.027) (0.020) (0.025) (0.020) LATE 4.220 6.735 ** 5.317 6.225 * (3.329) (3.136) (3.454) (3.218) 15564 15117 15250 15431 :(1) 1980 10, ImbensandKalyanaraman(2012) ;(2) LATE ( ) ( ) 3 BeckerandLewis(1973), ( ),,, 11, ( ) ( ),,,,,, 321, 19 1,,, 20 2 1979 1981 GDP ( ), 21 3 1979 1981 ( ),
918 ( ) 17 11,, RDD, ;, LATE,,Q-Q,,,, Q-Q, ( ),,,,,,Q- Q,,,, -, -,, 20 70,,,,,, (RDD), ( ) RDD,,
3 : 919,, :,,,,,, -,,, : - ;,,,,,,, ;,, [1] Angrist,J.D.,andA.B.Krueger, DoesCompulsorySchoolAtendanceAfectSchoolingand Earnings Quarterly Journalof Economics,1991,106(4),979 1014. [2] Angrist,J.D.,andV.Lavy, Using Maimonides RuletoEstimatetheEfectofClassSizeon ScholasticAchievement,Quarterly Journalof Economics,1999,114(2),553 575. [3] Angrist,J.D.,V.Lavy,andA.Schlosser, MultipleExperimentsfortheCausalLinkbetween thequantityandqualityofchildren,journalof Labor Economics,2010,28(4),773 824. [4] Banister,J.,China schanging Population.Stanford,CA:StanfordUniversityPress,1987. [5] Becker,G.S.,and H.G.Lewis, OntheInteractionbetweenthe Quantityand Qualityof Children,Journalof PoliticalEconomy,1973,82(2),279 288. [6] Black,S.E.,P.J.Devereux,andK.G.Salvanes, TheMoretheMerrier TheEfectofFamily SizeandBirthOrderonChildren seducation,quarterly Journalof Economics,2005,120(2), 669 700. [7] Buckles,K.S.,andD.M.Hungerman, SeasonofBirthandLaterOutcomes:OldQuestions, New Answers,Review of Economicsand Statistics,2013,95(3),711 724. [8] Cameron,L.,N.Erkal,L.Gangadharan,andX.Meng, LitleEmperors:BehavioralImpactsof China sone-childpolicy,science,2013,340(6130),272 273.
920 ( ) 17 [9] Chen,Y.,A.Ebenstein,M.Greenstone,and H.Li, EvidenceontheImpactofSustained ExposuretoAirPolutiononLifeExpectancyfrom China shuairiverpolicy,pnas,2013,110 (32),12936 12941. [10]Cooney,R.S.,J.Wei,and M.G.Powers, The One-Child Certificatein HebeiProvince, China:Acceptanceand Consequence,1978 1988,Population Researchand Policy Review, 1991,10(2),137 155. [11]DasGupta,M., ExplainingAsia s Missing Women :A New LookattheData,Population and DevelopmentReview,2005,31(3),529 535. [12]Ebenstein,A., The MissingGirls ofchinaandtheunintendedconsequencesoftheone-child Policy,Journalof Human Resources,2010,45 (1),87 115. [13]Fan,J.,andI.Gijbels,Local Polynomial Modeling and Its Applications.London,New York and Melbourne:ChapmanandHal,1996. [14]Glick,P.J.,A.Marini,andD.E.Sahn, EstimatingtheConsequencesofChangesinFertility onchildhealthandeducationinromania:ananalysisusingtwinsdata,corneluniversity, FoodandNutritioPolicyProgram WorkingPaperNo.183,2005. [15],,,2003 5, 1 10 [16]Hainmueler,J., EntropyBalancing:A MultivariateReweighting MethodtoProduceBalanced SamplesinObservationalStudies,Political Analysis,2012,20(1),25 46. [17]Haton,T.J.,andR.M.Martin, FertilityDeclineandtheHeightsofChildreninBritain,1886 1938,Explorationsin Economic History,2010,47(4),505 519. [18],,,2013 7, 66 77 [19]Imbens,G.,andK.Kalyanaraman, OptimalBandwidthChoicefortheRegressionDiscontinuity Estimator,Review of EconomicStudies,2012,79(3),933 959. [20]Imbens,G.,andT.Lemieux, RegressionDiscontinuityDesigns:A GuidetoPractice,Journal of Econometrics,2008,142(2),615 635. [21]Lee,D.S.,and T.Lemieux, Regression Discontinuity Designsin Economics,Journalof Economic Literature,2010,48(2),281 355. [22],, ( ),2010 4, 1539 1558 [23]Li,H.,J.Zhang,and Y.Zhu, The Quantity-Quality Trade-OfofChildreninaDeveloping Country:IdentificationUsingChineseTwins,Demography,2008,45(1),223 243. [24]Li,H.,J.Yi,andJ.Zhang, EstimatingtheEfectoftheOne-ChildPolicyontheSex Ratio ImbalanceinChina:IdentificationBasedontheDiference-in-Diferences,Demography,2011,48 (4),1535 1557. [25]Li,J.,andR.S.Cooney, SonPreferenceandtheOneChildPolicyin China:1979 1988, Population Researchand Policy Review,1993,12(3),277 296. [26]Li,J., China sone-childpolicy:howand How WelHasIt Worked A CaseStudyofHebei Province,1979 1988,Populationand DevelopmentReview,1995,21(3),563 585. [27], :,,2014 3, 11 44
3 : 921 [28]Liu,H., TheQuality-QuantityTrade-of:EvidencefromtheRelaxationofChina sone-child Policy,Journalof Population Economics,2014,27(2),565 602. [29],, ( ),2008 4, 1271 1300 [30]Lordan,G.,andP.Frijters, UnplannedPregnancyandtheImpactonSiblingHealthOutcomes, Health Economics,2013,22(8),903 914. [31],,, 2010 5, 46 49 [32]McCrary,J., ManipulationoftheRunning VariableintheRegressionDiscontinuityDesign:A DensityTest,Journalof Econometrics,2008,142(2),698 714. [33]Park,C.,and W.Chung, Sib-ShipSize,BirthOrder,andChildren seducationindeveloping Countries:Evidencefrom Bangladesh,HitotsubashiJournalof Economics,2012,53(1),1 23. [34]Qian,N., Quantity-QualityandtheOneChildPolicy:TheOnly-ChildDisadvantageinSchool EnrolmentinRuralChina,NBER WorkingPaperNo.14973,2009. [35]Rosenzweig,M.R.,andJ.Zhang, DoPopulationControlPoliciesInduceMoreHumanCapital Twins,Birth WeightandChina s One-Child Policy,Review of EconomicStudies,2009,76 (3),1149 1174. [36]Smith,J.P., UnravelingtheSES-HealthConnection,Populationand DevelopmentReview, 2004,30,108 132. [37]VanderKlaauw,W., EstimatingtheEfectofFinancialAid OfersonColegeEnrolment:A Regression-DiscontinuityApproach,InternationalEconomic Review,2002,43(4),1249 1287. [38] Wilis,R.J., A New ApproachtotheEconomic TheoryofFertility Behavior,Journalof PoliticalEconomy,1973,81(2),S14 S64. [39],,,2010 [40],, ( ),2011 4, 1187 1208 [41],,,2003 [42],,,2011 2, 125 131 [43]Zeng,Y.,P.Tu,B.Gu,Y.Xu,B.Li,and Y.Li, CausesandImplicationsoftheRecent IncreaseintheReportedSexRatioatBirthinChina,Populationand DevelopmentReview,1993, 19(2),283 302. [44]Zhang,J.,andB.G.Spencer, WhoSignsChina sone-childcertificate,and Why Journalof Population Economics,1992,5(3),203 215. [45],,,2011 10, 61 73
922 ( ) 17 TheImpactofBirthControlPolicy onoffspring seducationlevel: MicroeconomicEvidencefrom China XUEZHENG QIN RUDAI YANG (Peking University) CASTIELCHEN ZHUANG 1* (Universityof Washington) Abstract Theefectoftherecent two-child policy hasatractedanew waveof atention.usingthe2005 China sinter-census1% populationsurveydataand multiple econometricmethods,weidentifythecausalrelationshipbetweenquantityandqualityof populationin China. The results suggest that being a single child increased his/her educationalatainment,butthisefectwasonlyevidentinregionswithlowerincomeandless developedcredit markets. With China s economic development,relaxing households constraintsoninvestmentinofspring shumancapitalmayeasethenegativeimpactsofa country-wide relaxation of birth control and thus may promote sustainable economic development. Key Words birthcontrol,educationlevel,regressiondiscontinuitydesign JELClasification J13,D13,I20 * CorrespondingAuthor:CastielChenZhuang,UW 1 Economics,Savery HalRoom305,Seatle,WA 98195,UnitedStates.Tel:1-206-487-6429;E-mail:zogcee@gmail.com,cczhuang@uw.ed