Predicting Changes in Wealth Using the Cube Root of the Changes in Wealth as the Dependent Variable Vartanian: SW 541 libname in1 v6 'h:\general PSID'; data a;set in1.w94b; if divorce=1 or separat=1 then div=1;else div=0; if chwlth2<0 then negative=1;else negative=0; if chwlth2<0 then chwlth2=-chwlth2; if chwlth2=0 then chwlth2=1; cbrt=chwlth2**(1/3); if negative=1 then cbrt=-cbrt; tothrs2=tothrs2/1000; *1994 values for initial wealth; if wealth1<5000 then initial1=1;else if wealth1=. then initial1=.;else initial1=0; if 5000<=wealth1<22420 then initial2=1;else if wealth1=. then initial2=.;else initial2=0; if 22420<=wealth1<95760 then initial3=1;else if wealth1=. then initial3=.;else initial3=0; if 95760<=wealth1<284480 then initial4=1;else if wealth1=. then initial4=.;else initial4=0; if 284480<=wealth1 then initial5=1;else if wealth1=. then initial5=.;else initial5=0; *1994 housing values; if hsval94=0 then hs1=1;else hs1=0; if 0<hsval94<=54880 then hs2=1;else hs2=0; if 54880<hsval94<=117600 then hs3=1;else hs3=0; if 117600<hsval94<=221760 then hs4=1;else hs4=0; if hsval94>221760 then hs5=1;else hs5=0; if 0<youngest<6 then young=1;else if youngest=. then young=.;else young=0; if age>=65 then old=1;else old=0; if bigcity=. or urban=. then small=.;else if bigcity=0 and urban=0 then small=1;else small=0; if ne=1 or nc=1 or west=1 then north=1;else if ne=. or nc=. or west=. then north=.;else north=0; if if if if if if -10000<=fmns<=1 then inpov=1;else if fmns=. then inpov=.;else inpov=0; 1<fmns<=1.5 then nrpov=1;else if fmns=. then nrpov=.;else nrpov=0; 1.5<fmns<=2 then pov2=1;else if fmns=. then pov2=.;else pov2=0; 2<fmns<=3 then pov3=1;else if fmns=. then pov3=.;else pov3=0; 3<fmns<=5 then pov4=1;else if fmns=. then pov4=.;else pov4=0; fmns>5 then pov5=1;else if fmns=. then pov5=.;else pov5=0; C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 1 if nbpov ne . and rural ne . and femhead ne . and married ne . and dropout ne . and ftwork ne . and black ne . and clersal ne . and uncy ne .; proc reg data=a outest=a1; model cbrt=prin1 inpov nrpov pov2 pov3 pov4 black othrace female kids married div widowed young cohabit initial1-initial4 stocks reales hs1-hs4 dropout hsg scoll age old uncy clersal crafop servlab ftwork bigcity urban south gotmarr tothrs2 gotdiv gotwid yrsmarr yrsdiv; weight weight; ods output parameterestimates=Female84; output out=y p=pred; run; data d;set a1; rename prin1=prin11 inpov=inpov1 nrpov=nrpov1 pov2=pov21 pov3=pov31 pov4=pov41 initial1=initia11 initial2=initia21 initial3=initia31 initial4=initia41 black=black1 othrace=othrace1 female=female1 kids=kids1 married=married1 young=young1 stocks=stocks1 reales=reales1 hs1=hs11 hs2=hs21 hs3=hs31 hs4=hs41 dropout=dropout1 hsg=hsg1 scoll=scoll1 age=age1 old=old1 uncy=uncy1 clersal=clersal1 crafop=crafop1 servlab=servlab1 ftwork=ftwork1 bigcity=bigcity1 urban=urban1 south=south1 gotmarr=gotmarr1 tothrs2=tothrs21 gotdiv=gotdiv1 gotwid=gotwid1 yrsmarr=yrsmarr1 yrsdiv=yrsdiv1 div=div1 widowed=widow1; cloth=1; drop _type_; data e;set y;cloth=1; data f;merge d e;by cloth; xb=pred; xb_male=xb-female*female1; xb_fem=xb-female*female1+female1; xb_rich=pred-inpov*inpov1-nrpov*nrpov1-pov2*pov21-pov3*pov31-pov4*pov41; xb_inpov=xb_rich+inpov1; xb_nrpov=xb_rich+nrpov1; xb_pov2=xb_rich+pov21; xb_pov3=xb_rich+pov31; xb_pov4=xb_rich+pov41; xb_white=pred-black*black1-othrace*othrace1; xb_black=xb_white+black1; xb_othr=xb_white+othrace1; xb_in5=pred-initial1*initia11-initial2*initia21-initial3*initia31initial4*initia41; xb_in1=xb_in5+initia11; xb_in2=xb_in5+initia21; xb_in3=xb_in5+initia31; xb_in4=xb_in5+initia41; xb_hs5=pred-hs1*hs11-hs2*hs21-hs3*hs31-hs4*hs41; xb_hs1=xb_hs5+hs11; xb_hs2=xb_hs5+hs21; xb_hs3=xb_hs5+hs31; xb_hs4=xb_hs5+hs41; xb_nevM=xb-married*married1-gotmarr*gotmarr1-gotdiv*gotdiv1-gotwid*gotwid1yrsmarr*yrsmarr1yrsdiv*yrsdiv1-div*div1-widowed*widow1; xb_ngm=pred-gotmarr*gotmarr1-married*married1; C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 2 xb_gm=xb_ngm+gotmarr1; xb_marr=pred-married*married1-gotdiv*gotdiv1-gotwid*gotwid1-div*div1widowed*widow1+married1; xb_div=pred-gotdiv*gotdiv1-gotwid*gotwid1-div*div1-widowed*widow1married*married1+div1; xb_wid=pred-gotdiv*gotdiv1-gotwid*gotwid1-div*div1-widowed*widow1married*married1+widow1; xb_ngdw=pred-gotdiv*gotdiv1-gotwid*gotwid1-married*married1-div*div1widowed*widow1; xb_gd=xb_ngdw+gotdiv1; xb_gw=xb_ngdw+gotwid1; xb_coll=xb-dropout*dropout1-hsg*hsg1-scoll*scoll1; xb_drop=xb_coll+dropout1; xb_hsg=xb_coll+hsg1; xb_scoll=xb_coll+scoll1; xb_prof=xb-clersal*clersal1-crafop*crafop1-servlab*servlab1; xb_cler=xb_prof+clersal1; xb_craf=xb_prof+crafop1; xb_serv=xb_prof+servlab1; xb_rur=xb-bigcity*bigcity1-urban*urban1; xb_big=xb_rur+bigcity1; xb_urb=xb_rur+urban1; xb_south=xb-south*south1+south1; xb_north=xb-south*south1; xb_stock=xb-stocks*stocks1+stocks1; xb_real=xb-reales*reales1+reales1; if female=1 then do; xb_kidl=xb-kids*kids1+kids1*0; xb_kidh=xb-kids*kids1+kids1*2.5; xb_ymarrl=xb-yrsmarr*yrsmarr1+yrsmarr1*26.52; xb_ymarrh=xb-yrsmarr*yrsmarr1+yrsmarr1*0; xb_ydivl=xb-yrsdiv*yrsdiv1+yrsdiv1*0; xb_ydivh=xb-yrsdiv*yrsdiv1+yrsdiv1*1.54; xb_ageh=xb-age*age1+age1*56.63-old*old1; xb_agel=xb-age*age1+age1*27.5-old*old1; xb_prinl=xb-prin1*prin11+prin11*(-2.05); xb_prinh=xb-prin1*prin11+prin11*(2.21); xb_unl=xb-uncy*uncy1+uncy1*4.76; xb_unh=xb-uncy*uncy1+uncy1*6.46; xb_tothl=xb-tothrs2*tothrs21+tothrs21*0.51; xb_tothh=xb-tothrs2*tothrs21+tothrs21*3.14; xb_FTL=xb-ftwork*ftwork1+ftwork1*0; xb_FTH=xb-ftwork*ftwork1+ftwork1*7.42; end; if female=0 then do; xb_kidl=xb-kids*kids1+kids1*0; xb_kidh=xb-kids*kids1+kids1*1.94; xb_ymarrl=xb-yrsmarr*yrsmarr1+yrsmarr1*0; xb_ymarrh=xb-yrsmarr*yrsmarr1+yrsmarr1*27.28; xb_ydivl=xb-yrsdiv*yrsdiv1+yrsdiv1*0; xb_ydivh=xb-yrsdiv*yrsdiv1+yrsdiv1*1.59; xb_ageh=xb-age*age1+age1*53.77-old*old1; xb_agel=xb-age*age1+age1*27.40-old*old1; xb_prinl=xb-prin1*prin11+prin11*(-2.05); xb_prinh=xb-prin1*prin11+prin11*(2.33); C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 3 xb_unl=xb-uncy*uncy1+uncy1*4.83; xb_unh=xb-uncy*uncy1+uncy1*6.64; xb_tothl=xb-tothrs2*tothrs21+tothrs21*1.04; xb_tothh=xb-tothrs2*tothrs21+tothrs21*3.61; xb_FTL=xb-ftwork*ftwork1+ftwork1*.24; xb_FTH=xb-ftwork*ftwork1+ftwork1*7.93; end; pred1=pred; rich=(xb_rich)**3; inpov=(xb_inpov)**3; nrpov=(xb_nrpov)**3; pov2=(xb_pov2)**3; pov3=(xb_pov3)**3; pov4=(xb_pov4)**3; white=(xb_white)**3; black=(xb_black)**3; othrace=(xb_othr)**3; pred=(pred)**3; initial1=(xb_in1)**3; initial2=(xb_in2)**3; initial3=(xb_in3)**3; initial4=(xb_in4)**3; initial5=(xb_in5)**3; hs1=(xb_hs1)**3; hs2=(xb_hs2)**3; hs3=(xb_hs3)**3; hs4=(xb_hs4)**3; hs5=(xb_hs5)**3; gotmarr=(xb_gm)**3; married=(xb_marr)**3; Div=(xb_div)**3; Wid=(xb_div)**3; gotdiv=(xb_gd)**3; gotwid=(xb_gw)**3;; dropout=(xb_drop)**3; hsg=(xb_hsg)**3; scoll=(xb_scoll)**3; college=(xb_coll)**3; kidl=(xb_kidl)**3; kidh=(xb_kidh)**3; ymarrl=(xb_ymarrl)**3; ymarrh=(xb_ymarrh)**3; ydivl=(xb_ydivl)**3; ydivh=(xb_ydivh)**3; agel=(xb_agel)**3; ageh=(xb_ageh)**3; prinl=(xb_prinl)**3; prinh=(xb_prinh)**3; stock=(xb_stock)**3; real=(xb_real)**3; unl=(xb_unl)**3; unh=(xb_unh)**3; tothl=(xb_tothl)**3; tothh=(xb_tothh)**3; prof=(xb_prof)**3; cler=(xb_cler)**3; C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 4 craf=(xb_craf)**3; serv=(xb_serv)**3; rur=(xb_rur)**3; big=(xb_big)**3; urb=(xb_urb)**3; south=(xb_south)**3; north=(xb_north)**3; ftworkL=(xb_ftl)**3; ftworkH=(xb_fth)**3; nevmarr=(xb_nevm)**3; Female=(xb_fem)**3; Male=(xb_male)**3; predict=pred; proc means data=f;var predict inpov nrpov pov2-pov4 rich initial1-initial5 white black othrace hs1-hs5 gotmarr married div wid gotdiv gotwid dropout hsg scoll college kidl kidh ymarrl ymarrh ydivl ydivh agel ageh prinl prinh stock real unl unh tothl tothh prof cler craf serv rur big urb south north ftworkl ftworkh nevmarr male female; weight weight; output out=aaa; run; C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 5 The SAS System 17:10 Monday, February 3, 2003 101 The REG Procedure Model: MODEL1 Dependent Variable: cbrt Weight: WEIGHT Analysis of Variance DF Sum of Squares Mean Square 44 4831 4875 1211653 9355533 10567186 27538 1936.56247 Root MSE Dependent Mean Coeff Var 44.00639 12.02779 365.87272 Source Model Error Corrected Total R-Square Adj R-Sq F Value Pr > F 14.22 <.0001 0.1147 0.1066 Parameter Estimates Variable Intercept Prin1 inpov nrpov pov2 pov3 pov4 BLACK OTHRACE FEMALE KIDS MARRIED div WIDOWED young COHABIT initial1 initial2 initial3 initial4 STOCKS REALES hs1 hs2 hs3 hs4 DROPOUT DF Parameter Estimate Standard Error t Value Pr > |t| 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 13.07256 0.38123 -23.14475 -20.69407 -23.59939 -20.56332 -15.28790 -9.62956 -7.08881 -2.62394 0.53898 6.85406 -1.45770 6.18941 -1.01361 -11.00013 40.61934 34.41435 27.79630 16.49662 3.45450 -2.15505 -6.71834 -7.39262 -7.22473 -1.29228 -5.21302 8.02565 0.46147 3.65036 3.32698 3.05000 2.32053 1.94478 2.35975 4.63767 1.39393 0.75887 3.11398 2.70144 3.81949 1.84752 5.06129 3.07804 2.92291 2.48615 2.45867 1.59583 1.81024 2.78224 3.05884 2.51040 2.38867 2.33081 1.63 0.83 -6.34 -6.22 -7.74 -8.86 -7.86 -4.08 -1.53 -1.88 0.71 2.20 -0.54 1.62 -0.55 -2.17 13.20 11.77 11.18 6.71 2.16 -1.19 -2.41 -2.42 -2.88 -0.54 -2.24 0.1034 0.4088 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.1264 0.0598 0.4776 0.0278 0.5895 0.1052 0.5833 0.0298 <.0001 <.0001 <.0001 <.0001 0.0305 0.2339 0.0158 0.0157 0.0040 0.5885 0.0254 C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 6 The SAS System 17:10 Monday, February 3, 2003 102 The REG Procedure Model: MODEL1 Dependent Variable: cbrt Parameter Estimates Variable DF Parameter Estimate Standard Error t Value Pr > |t| HSG SCOLL AGE old UNCY CLERSAL CRAFOP SERVLAB FTWORK BIGCITY URBAN SOUTH GOTMARR TOTHRS2 GOTDIV GOTWID YRSMARR YRSDIV 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1.39953 -0.04662 0.06938 -15.42974 -2.27268 -5.73311 -3.71115 -2.64253 0.23883 5.48570 1.81425 -3.47353 8.28298 0.47443 -24.41912 -3.09902 0.02253 0.08345 1.98033 1.92487 0.11668 2.87024 0.80700 2.04659 1.80008 2.83549 0.20875 1.85917 1.54526 1.39245 2.85034 0.75324 3.04051 3.96882 0.09797 0.46948 0.71 -0.02 0.59 -5.38 -2.82 -2.80 -2.06 -0.93 1.14 2.95 1.17 -2.49 2.91 0.63 -8.03 -0.78 0.23 0.18 0.4798 0.9807 0.5521 <.0001 0.0049 0.0051 0.0393 0.3514 0.2526 0.0032 0.2404 0.0126 0.0037 0.5288 <.0001 0.4349 0.8181 0.8589 C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 7 The SAS System 17:10 Monday, February 3, 2003 103 The MEANS Procedure Variable N Mean Std Dev Minimum Maximum ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ predict 4876 10472.90 22937.00 -191590.90 253980.69 inpov 4876 329.1510027 9751.41 -218497.99 64886.78 nrpov 4876 1893.11 9522.74 -192897.46 77497.19 pov2 4876 53.6538467 9876.19 -223483.28 62709.24 pov3 4876 1981.32 9532.43 -191590.90 78212.37 pov4 4876 6167.00 11733.54 -143660.16 110873.88 rich 4876 31711.22 32912.80 -51005.61 253980.69 initial1 4876 42320.56 58294.06 -37683.85 312419.68 initial2 4876 26755.93 41553.08 -62718.65 234310.36 initial3 4876 15143.74 27780.28 -99570.29 166661.72 initial4 4876 2334.66 15196.06 -191590.90 83636.37 initial5 4876 -16252.62 29072.04 -407620.68 20202.06 WHITE 4876 11193.15 23306.63 -110726.29 253980.69 BLACK 4876 2184.11 14451.27 -191590.90 154846.29 OTHRACE 4876 4143.85 15754.31 -167359.13 177882.10 hs1 4876 7675.36 17780.40 -186586.34 181420.57 hs2 4876 7006.39 17087.29 -193269.69 175014.85 hs3 4876 7170.41 17255.58 -191590.90 176595.50 hs4 4876 14247.04 24894.41 -138321.03 238747.63 hs5 4876 16188.02 26962.87 -128208.95 253980.69 GOTMARR 4876 13487.34 25504.67 -120305.01 271563.72 MARRIED 4876 13460.46 25271.66 -63247.68 253980.69 div 4876 4645.87 15551.25 -111662.68 166528.38 Wid 4876 4645.87 15551.25 -111662.68 166528.38 GOTDIV 4876 -14989.57 30358.52 -359665.55 32938.70 GOTWID 4876 3360.80 14479.22 -123474.21 152064.42 DROPOUT 4876 5414.84 16876.75 -247855.20 196753.99 HSG 4876 12466.61 24800.99 -177531.12 271780.51 SCOLL 4876 10670.39 22719.76 -191590.90 253980.69 COLLEGE 4876 10725.74 22784.00 -191126.46 254542.02 kidl 4876 9960.01 22351.56 -202540.63 245432.07 kidh 4876 11367.96 23931.27 -188916.57 257935.14 ymarrl 4876 10303.57 22706.78 -186575.56 246996.32 ymarrh 4876 10373.84 23030.49 -192491.02 254327.90 ydivl 4876 10465.93 22926.70 -191590.90 253980.69 ydivh 4876 10620.76 23107.13 -190312.40 255580.52 agel 4876 10330.05 21399.26 -192630.45 237967.34 ageh 4876 12665.45 24093.82 -173105.59 259674.35 prinl 4876 9184.10 21118.62 -198876.74 240316.84 prinh 4876 11092.09 23322.90 -182734.24 260204.63 STOCK 4876 12614.72 24839.29 -159171.45 253980.69 real 4876 8516.51 20515.30 -213890.41 253980.69 unl 4876 12229.24 24929.72 -170371.46 281389.97 unh 4876 7663.85 19654.38 -208532.74 231654.80 tothl 4876 9279.27 21140.84 -201336.63 238289.54 tothh 4876 10686.08 22765.47 -188749.79 252627.05 prof 4876 12316.90 24327.02 -139926.65 253980.69 cler 4876 5964.92 17497.21 -191590.90 191058.60 ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 8 The SAS System 17:10 Monday, February 3, 2003 104 The MEANS Procedure Variable N Mean Std Dev Minimum Maximum ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ craf 4876 7965.89 19526.09 -172130.34 211895.27 serv 4876 9119.99 20772.42 -162399.60 223495.13 rur 4876 8901.45 21036.03 -251654.06 193531.35 BIG 4876 15983.75 29167.37 -191590.90 253980.69 urb 4876 10982.37 23423.69 -230576.76 212318.90 SOUTH 4876 7840.23 19590.86 -191590.90 214439.12 north 4876 11773.96 24118.42 -159003.87 253980.69 ftworkL 4876 9441.02 21933.14 -198823.71 236870.57 ftworkH 4876 11585.26 24419.73 -181257.47 258596.45 NEVMARR 4876 4873.85 15502.34 -109685.73 174570.81 Male 4876 12009.19 24318.97 -166602.37 253980.69 FEMALE 4876 8923.00 20761.18 -191590.90 223700.65 ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ C:\WP60\LECT2.PHD\NonLinear\cuberoot.wealth.doc 9