ࡱ> FHE~q` hbjbjqPqP .x::cAC     """6~~~8R$6"444$hn~"""44"4"4""4v B~0p"^666$ Z$666Z666"""""" Simultaneous Equation Regression and Causation Isolation and Causation y1 = g11x1 Variable x1 is the isolated  cause of adjustments of y1. y1 = g11x1+z1 There is lack of isolation because z1 is an unobserved other  cause. Pseudo-isolation: x1 and z1 are independent causes. What if the causes are not independent? For example suppose that x also cause y2 to change and this influences y1, too, as seen below. y1 = g11x1+b12y2+z1 where y2=g21x1+z2.  Example: let x1 be the education level of a worker, y1 be the number of on-the-job mistakes and y2 be boredom. Higher education may induce boredom on the job and at the same time directly reduce the number of mistakes, but boredom increases mistakes. a. What if we omit the intervening variable y2 and estimate y1= g11* x1+z1* using OLS? plim g11* = cov(y1,x1)/var(x1)= cov(g11x1+b12y2+z1,x1)/var(x1) = cov(g11x1+b12(g21x1+z2) +z1,x1)/var(x1)=(g11+b12g21)var(x1)/ var(x1) = g11 + b12 g21.  If g11 = - b12 g21, then plim g11*=0. This is a suppressor relationship. If g11=0 but b12g21>0 then plim g11*>0. This is a confounding relationship. b. What if we omit a common cause? In the above, suppose we do not measure x1; we might estimate y1 = b12* y2+z1*. This means that the error is z1*= g11x1+z1. plim b12*=cov(y1,y2)/var(y2)= cov(g11x1+b12y2+z1,y2)/var(y2)= b12+g11cov(x1,y2)/var(y2) = b12+g11 EMBED Equation.3 . If b12=0 but g11 EMBED Equation.3 `"0, then we infer that y2 causes y1 when this is false: a confound. a. ( Causation does not prove correlation b. ( Correlation does not prove causation Thus, simple correlation analysis can neither prove nor disprove causation. However, recent work has shown that complex correlation patterns can establish causation; see notes Do Wet Lawns Cause the Grass to Grow. Timing and Direction of Causation: Rather than use =, use ( to denote direction of causation. Truth: yt ( g xt+zt and xt(axt-1+nt. Note that (xt+1-nt+1)/a (xt and hence yt ( g xt+zt((g/a)xt+1+zt-(g/a)nt+1. That is, yt is correlated with xt+1 because of the common cause xt. If we were to ignore the direction of causation, focusing on post hoc, ergo propter hoc, one could easily find that a regression xt+1 = b*yt+zt+1* has a highly significant b* assuming a/g(0 and falsely believe that yt causes xt+1. It looks like yt is determined before xt+1 but actually they are determined simultaneously by xt. Simultaneous Equation Bias y1 = ay2 + bx1 +e1, Structural y2 = ay1 + bx2 +e2. Equations  Solve for the endogenous variables  EMBED Equation.3  Reduced Form Equations Look at just the first structural equation (a similar analysis holds for the second equation) and estimate it via OLS. Multiply it by y2 and then by x1 to get two normal equations: y2y1=ay22+by2x1+y2e1, x1y1=ax1y2+bx12+x1e1. Note: plim x1e1 = 0, but plim y2e1=plim[bx2e1+abx1e1+ae12+e1e2]/(1-aa)=as12/(1-aa)`"0. OLS estimators of a and b are therefore  EMBED Equation.3 . Focus on  EMBED Equation.3 ; a similar analysis applies to  EMBED Equation.3 . Substituting from the normal equations gives  EMBED Equation.3  Notice that the terms involving b cancel out. This can be expressed as  EMBED Equation.3 . Making use of the reduced forms, we have that plim[y2x1]=(bx2+abx1)x1/(1-aa) and plim[y22]=E[y22]=[(bx2+abx1)2+a2s12+s22]/(1-aa)2. Recall, plim xe1= 0 and plim[y2e1]=as12/(1-aa). Using the fact that the plim[g(X)]=g(plim[X]) for a continuous function g( ), plim EMBED Equation.3 . Simultaneous Equation Bias For illustration, if a>0 and aa<1, then plim EMBED Equation.3  > a, so the OLS estimate exceeds the true a. This is because y2 is correlated with e1. If e1 had a bump up, it would cause y1 to go up, and this (through the second equation) would cause y2 to go up (if a>0). Hence, OLS would attribute both bumps in y2 and the unobserved e1 to just the variable y2, and the estimate of the strength of y2, the parameter a, would seem larger than it is in truth. Numerical example: y1 = 0.1y2 + x1 +e1, y2 = 2y1 -3x2 +e2 and suppose that si2=1. When e1 goes up by 1 unit, y1 will rise by 1, and from the second equation, y2 will rise by 2 units. This in turn will cause y1 to increase by an additional 0.1(2=0.2 units (this feedback between the two equat@\^     " & ( , . 0 2 4 6 ~   B F J L P R T V X \ ^ ` b d f v x z | 潲jh #h:}U h B~h:}h B~h:}OJQJ hoW1h:}h!<h:}OJQJ h:}H*h:}hoh:}>* h:}5h)uh:}5E^ . .Jp.+j^gd:} 0^`0gd:} ^`gd:}gd:}$a$gd:}ggh * +     HJNR`bfhvx h B~h:}OJQJ h B~h:} h:}H*h B~h:}H*h!<h:}OJQJ h'h:} h:}H*h:} h #h:}K   "&248:DFNPVZ\bdhjlnpvx~   $&*DFJLj:>fjnpƿ h6h:}h6h:}OJQJh6h:}>* h:}H* hUh:}jh:}0JUhoW1h:}0J h B~h:}h B~h:}OJQJh!<h:}OJQJh:} h:}H*?ptvxz|~  (*.0246:<>@BDHJXZ`bfhjnxz~hoh:}OJQJh B~h:}OJQJ h'h:}h'h:}OJQJ hUh:}h!<h:}OJQJh:} h:}H* h:}H*F&(*,0\^pr-.5y(Ϲό|||uuphhoh:}>* h:}>* h.{h:} h:}5 jhoh:}5hoh:}5 hoh:}j5 hoh:}EHH*UjBC h:}CJUVaJhoh:}OJQJh B~h:}OJQJh:}j hoh:}EHH*UjLBC h:}CJUVaJ h:}H*jh:}H*U'()EFijstuv   "(*,.LRTV\`bdfhjPR jh:}hh:}OJQJhoh:}OJQJ jh:}CJ h:}H*hoh:}>* jh:}h:} hh:}GZbfhjlnprxz DFtz $&(24>@DFHJNPѲܫᤝ~ hPh:}h.h:}OJQJ h:}5h@mh:}5 h.{h:} h>h:} h2h:} jh:}h:}OJQJh2h:}OJQJ h:}H*hd9Qh:}OJQJ h:}H*h:}h!rFh:}56 h:}56hd9Qh:}561"$fRT %&(!`gd:}$a$gd:}gd:}Pdfhjprtv|~$PRdfѷ򨙷 h:}H* h'uh:}h.h:}H*jh'uh:}EHH*UjeGC h:}CJUVaJjh:}H*UjXhPh:}U h:}5 hPh:}h.h:}OJQJh'uh:}OJQJh:} h:}H*hPh:}54  468:<Z\^`lnprtvxz~   !jfGC h:}CJUVaJjh:}Uh yh:}OJQJhPh:}OJQJ hPh:}h'uh:}OJQJ h:}H* h:}H* h'uh:} hUh:}h:}=!"#$&/0CDEFfgz{|}#$%&\]^_     涩暍~qfh5!h:}OJQJjI-hFWh:}EHUjugGC h:}CJUVaJjV(h!h:}EHUjgGC h:}CJUVaJj[&h!rFh:}EHUj1mGC h:}CJUVaJj #h5!h:}EHUjgGC h:}CJUVaJ h:}H*h:}jh:}Ujh5!h:}EHU%     " : < > H J L T V X Z \ ^ b d f h j l n p r t v x z | !!!!!!!"񯢾h,h:}5jc0hFWh:}EHUjdbGC h:}CJUVaJjh:}Uh yh:}OJQJh'uh:}OJQJh5!h:}OJQJ h:}H* h:}H*h:}hFWh:}OJQJ:! " "%%LLLLQQQQQ8TdTTUUUUULX~X4YYY`gd:}$a$gd:}gd:}$a$gd:}" "6"8"H"J"d"f"""""##<#>#@#L#N#P#d######$$.$0$$$$$$$$%<%>%%%%%%%%%%%&&&&& &&»»©h.h:}OJQJh #h:}>* h:}>* hMOh:}hMOh:}OJQJ h,h:} h:}H*j&3h,h:}EHUjxCC h:}CJUVaJjh:}Uh,h:}OJQJh:} h:}58&&&&& &$&&&(&L&N&P&R&f&h&j&&&&&'Z'\'''(JJJ K KLLLLLLLLNNNNOO~jJ h:}CJUVaJhoW1h:}OJQJhDh:}5j05hFWh:}EHUjCC h:}CJUVaJjh:}UU jh:} hkh:} h:}H*hkh:}OJQJh.h:}OJQJ h:}H*h:}h:}OJQJ,ions would continue, but I will stop here for pedagogical purposes). Empirically, we will have observed y2 going up by 2 units and y1 going up by 1.2 units, which seems to imply that the coefficient  a has a value like 1.2/2=0.6, rather than the true 0.1. Including all the feedbacks, the value we would expect from OLS is  EMBED Equation.3 . Conclusion: If an endogenous variable is part of a simultaneous equation system with feedback loops between this and the other endogenous variables, than traditional regression will not reveal the truth on average. (Note: if a=0, there is no feedback loop and plim EMBED Equation.3 .) That is, the OLS estimators are asymptotically biased estimates of the true coefficients in the equations. We must therefore do something other than OLS to deal with the simultaneous equation bias, and this must take into account the fact that there are other equations determining some of the explanatory variables. Two Stage Least Squares The problem with OLS in a simultaneous equation model is that the errors are correlated with the regressors (in the above y2 was correlated with e1). In creating the estimators of the first equation via OLS, we multiplied the first structural equation by both its regressors (y2 and x1) to get the normal equations: y2y1=ay22+by2x1+y2e1, x1y1=ax1y2+bx12+x1e1. Instead of doing this, suppose that we multiplied the first structural equation by the two exogenous variables, x1 and x2: x2y1=ax2y2+bx2x1+x2e1 x1y1=ax1y2+bx12+x1e1. Notice that plim x2e1=0 because x2 is exogenous. The variable x2 is an instrumental variable, since it is both i) causally linked to y2 and ii) independent of e1. Hence if we took the plim of these normal equations, the terms involving errors would drop out and we could solve for estimators of a and b as  EMBED Equation.3 . Again, focus on  EMBED Equation.3 and substitute from the new normal equations to get  EMBED Equation.3 Taking the plim of both sides gives  EMBED Equation.3 . This procedure is called two-stage least-squares because it can be accomplished slightly differently. First stage, run an auxiliary regression of the reduced form equation: y2=h1x1+h2x2+q2. Given the OLS estimates of this reduced form equation coefficients, compute the predicted values of the endogenous variable EMBED Equation.3 . Second stage, run an OLS regression for the structural equation using not the original y2 variables (that are correlated with errors), but the predicted values  EMBED Equation.3 from the first stage regression:  EMBED Equation.3 . Since the predicted values are just a weighted average of the exogenous variables, both of the explanatory variables in this second stage regression are independent of the errors, and the second stage regression will give an unbiased estimate of  a and  b. In essence, we have used a weighted average of the exogenous variables as an instrumental variable, rather than the other exogenous variable (as seen above). In either case, the outcome is the same. Finally, the computation of standard errors of the coefficients should be based upon the standard error of the regression, but this should be calculated from deviations of y1 from the predicted value  EMBED Equation.3 , not  EMBED Equation.3 . Notice that in both the 2SLS coefficients are used, but in the former the observed value of y2 is used while in the latter the first stage predicted value  EMBED Equation.3  is used. We need to use the observed value y2 or else the deviation is not a measure of the true error. When you run the second stage regression, the printed standard errors are inappropriately too small.   plim is explained in notes on  Convergence in Probability.     James Hess October 18, 2007 PAGE  PAGE 1 OOOQQQRRRRRRRSSSS:TT@TFTHTJTPTRTTTVTZT\T^T`TbTdTfThTjTlTrTtTvTxT~TTTTTTTrUtUUUUUUUUUUUUUԽԲ櫤ԽԲ h.h:} h'uh:} hUh:}h'uh:}OJQJ h:}H* hDh:}hDh:}OJQJ h:}H*hDh:}5 h:}5h:}jh:}Uj8hoW1h:}EHU>UUUUUUUUUUUUUUUUUUUUUUUUVV V V$V&VbVdVrVVVV"W$W&WLXNXtXvXxXzX|XXXXXjh:}U hmYh:}j:h.h:}EHH*Uj"CC h:}CJUVaJh.h:}H*jh.h:}H*Uh\h:}OJQJh\h:}5 h:}H* h'uh:}h'uh:}OJQJ h:}H*h:}1XXXX4Y6Y\Y^Y`YbYYYYYYYZ[@[B[D[F[H[J[L[N[P[R[T[V[R\T\z\Ŷךׅ׀xoh]׀]hh:}OJQJ hh:}h:}H*OJQJh:}OJQJ h:}H*h)uh:}>*j5FhFWh:}EHUjCC h:}CJUVaJjAhUh:}EHUj\CC h:}CJUVaJ hUh:}jhUh:}Uh:}jh:}Uj?hmYh:}EHUjCC h:}CJUVaJ#z\|\~\\]]6]8]]]]]]]6^8^^^`^b^d^^c`ctccccccccccc׻ןʋ|o`jeL h:}CJUVaJjOhh:}EHUj@L h:}CJUVaJ h\lh:}jMhh:}EHUj]CC h:}CJUVaJj~Khh:}EHUjkGC h:}CJUVaJ h:}H*h)uh:}>*h:}jh:}UjIhh:}EHUj޾CC h:}CJUVaJ cccdd8e:e`ebedefeeeggg ggggggggggggggggggggghhhh h hhh澷橡}r}h:}0JmHnHuhh!rF h!rF0Jjh!rF0JUhMhhy}jh|BUh|Bjh:}0JU h:}h h\lh:}jThh:}EHUjkGC h:}CJUVaJ h:}H*h:}jh:}UjDRhh:}EHU,Yggggggggggggggggg h hhhh &`#$gdgd:}gd:}`gd:}hh h:}h,1h/ =!"#$%  Dd p CE0  # A" O|P- = >Ej D`@=b O|P- = >EԈ94f,]0 xZ L~`#8nb8N&rUb"? WvA1`+ J]FXnX8bƠ:,)8URT%m vw,,cYpξoޛ=~{0ufj@Ά-Ew\&, S @ ^aԽb;wS.aLشM;S)x8GR! i;X?X[,Z@mL)mkIIw"~~tӚXV-W[C^(0Y0SU[Bq-uZlVBYP,^j [l1x~~˚j*E jxϪV$_.S!x[%Ks8pY:hJoci*\Ňʚ*k䆕ỳAXA i][ڣ.)eЫxĝjYrb)3F;3V*݌+%7c\pV.Wk*c y6$bYF'U{˩nye,[bIj#rTfy*--D]*=j,PU uD>S-Bn *᲼B /iry`ot~w_Ի̪D Qs~O+ty?<ӭYPjym@DEjm(VFU*Ux7.2Jx}B'Br\3J_,9 !c5q{ۺ^Qhy{7ݼlĩu/F;vT_-FGhk1eq2~E~=_<ŇٶCe,MfbAA ᡅG194P 5P [,y'@~LN[w?\+ҙqW%ݷ%j05(ޣ xt~q:tg iq1Kz5a^ՏF7.A bەy';FL+t@ôv9nC*t9qn@\O1Ww&g捦q$&lFKm5Yw 22 %Z7O`Q>88]qx f@:JJ4 sfc mx? OuB*ێ|5q#$+͇4_v~g!C>x1X`CCn!1#hj38sԆ3ϡi氮;|fKycb>DZgӐdh 1p"MJh 5g6=1v1CBb {,ٌNSfO'v(<73ܛ]Zfn|>|V=Ď4e>c' s')L_++1]v驍|TOvC뉁 X' |1?yَ{뢢'EԓŽO?8 =vF6@2XVܭV|iՊ:4©uL2 7JqsH)8s`T8ZLd߳k|8|t_૭ V8yYs|]XHfZv(*ZpjE/"ăկ'~E`8żkݷ5ٓK>AԾ"ӚERsi8>%ՙB%YA#ЙƔ|aH`hљ|/}92\7px]flAƾf}4~Nۼ)㡩.;[i;r{=}. ѓ8QKhg<027083KFƗ1}ogrP>Ʈ\0?1QMz0]tG`>:>yԊՅFTatkI3 *iJ|xL<- zY`"użmx}-4fCfX7W JJ3Q־^#+hwI{#؟ЂE߫*:&*,Z(8 񹥘"~~" [;ߨFN Zl69y6?;c0a~.5̇6C3WӍk3|8"/+4=.rtl  {-$.'nlZQe(jLEY/fَHQ˟3lw"Lof>ت'z{b>x1*zq!zQx)zDz?|5pzcT4zQ(3xS}ȇvzpC1Ǜ%)TS}25w/ֲ/m(?Y#Dd |b  c $A? ?3"`?2m1_*]@\=ަIV ``!A1_*]@\=ަ `0xcdd``fed``baV d,FYzP1n:&f! KA?H1Zjـ7$# lo`'0LY ZZǰ@pT9 P.P56`ci5{7@`ܞVVR`~Ԟb#ԞF+I\P,Xh 0y{qĤ\Y\ q1(2t)v<^'f~ra4#Dd |b  c $A? ?3"`?2m x7~Y586ַ2Iy ``!A x7~Y586ַ2 `0xcdd``fed``baV d,FYzP1n:&f! KA?H1Zjـ7$# lo`'0LY ZZǰ@pT9 P.P56`ci5{7@`ܞVVR`~Ԟb#ԞF+I\P,Xh 0y{qĤ\Y\ q1(2t/v<^'f~}a~ Dd I\ 0  # A" D-cAbrI `  !"#$%&'()*+,-./0123456789:;<>?@ABCDGJiKLNMPOQSRUTWVYX\Z[]^`_abcdefgjhlkmnoqprstuvwyxz{|}Root Entryn FBI7Data =VWordDocumentm.xObjectPoolp `BB_1128432972F `B `BOle CompObjfObjInfo  !"#$%&'()*-0123458=@ABCDEFGHIJKLMNOPSVWXYZ[^abcfknopqrux{~ FMicrosoft Equation 3.0 DS Equation Equation.39q:D b x 1 y 2  FMicrosoft Equation 3.0 DS EqEquation Native V_11284330261 F `B `BOle CompObj fuation Equation.39q:D b x 1 y 2  FMicrosoft Equation 3.0 DS Equation Equation.39qObjInfo Equation Native  V_1128752555F `B `BOle  CompObj fObjInfoEquation Native ^_11287528426"F `B `BBKD y 1 y 2 []=1"a"1[] "1 bx 1 + 1 x 2 + 2 []=11"abx 1 +ax 2 + 1 +a 2 x 2 +bx 1 + 2 + 1 [] FMicrosoft Equation 3.0 DS Equation Equation.39qzD 2a 2b []=y 2  2Ole CompObjfObjInfoEquation Native * y 2 x 1 x 1 y 2 x 1  2 [] "1 y 2 y 1 x 1 y 1 []=1(y 2  2 )(x 1  2 )"(y 2 x 1 ) 2 x 1  2 "y 2 x 1 "y 2 x 1 y 2  2 []y 2 y 1 x 1 y 1 [] FMicrosoft Equation 3.0 DS Equation Equation.39q|@ 2a =1(_1128752901F `B `BOle +CompObj,fObjInfo.Equation Native /_1128754481'cF `B `BOle 6CompObj 7fy 2  2 )(x 1  2 )"(y 2 x 1 ) 2 x 1  2 "y 2 x 1 []y 2 y 1 x 1 y 1 [] FMicrosoft Equation 3.0 DS Equation Equation.39qD 2b  FMicrosoft Equation 3.0 DS Equation Equation.39q,; 2a =1(ObjInfo!9Equation Native :-_1128752922$F `B `BOle ;CompObj#%<fObjInfo&>Equation Native ?H_1128753013Y)F `B `By 2  2 )(x 1  2 )"(y 2 x 1 ) 2 x 1  2 "y 2 x 1 []ay 2  2 +by 2 x 1 +y 2  1 ax 1 y 2 +bx 1  2 +x 1 []=1(y 2  2 )(x 1  2 )"(y 2 x 1 ) 2 a((y 2  2 )(x 1  2 )"(y 2 x 1 ) 2 )+(y 2  1 )(x 1  2 )"(x 1  1 )(y 2 x 1 )(). FMicrosoft Equation 3.0 DS Equation Equation.39qgB$ 2a =a+(y 2  1 )(x 1  2 )"(x 1  1 Ole QCompObj(*RfObjInfo+TEquation Native U)(y 2 x 1 )(y 2  2 )(x 1  2 )"(y 2 x 1 ) 2 FMicrosoft Equation 3.0 DS Equation Equation.39q_1128751716J.F `BBOle \CompObj-/]fObjInfo0_Equation Native `_1128495313E3FBBOle dCompObj24efzD [2a ]=a+(1"a) 1  2  2  1  2 + 2  2 FMicrosoft Equation 3.0 DS Equation Equation.39qObjInfo5gEquation Native h5_1128504717 ,8FBBOle i D [2a ] FMicrosoft Equation 3.0 DS Equation Equation.39q5KD a+(1"a) 1  2  2  1  2 + 2  CompObj79jfObjInfo:lEquation Native mQ_1254551445=FBB2 =0.1+20.8141+1=0.1+0.32=0.42 FMicrosoft Equation 3.0 DS Equation Equation.39q%P,] [2a ]=aOle sCompObj<>tfObjInfo?vEquation Native wA FMicrosoft Equation 3.0 DS Equation Equation.39q8P\ 2a2b[]=x 2 y 2 x 2 x 1 x 1 _1128504866BFBBOle yCompObjACzfObjInfoD|Equation Native }T_1128498305GFBBOle CompObjFHfy 2 x 1  2 [] "1 x 2 y 1 x 1 y 1 []=1(x 2 y 2 )(x 1  2 )"(x 2 x 1 )(x 1 y 2 )x 1  2 "x 2 x 1 "x 1 y 2 x 2 y 2 []x 2 y 1 x 1 y 1 [] FMicrosoft Equation 3.0 DS Equation Equation.39q@ 2aObjInfoIEquation Native -_1128504924@TLFBBOle  FMicrosoft Equation 3.0 DS Equation Equation.39q¤D 2a=1(x 2 y 2 )(x 1  2 )"(x 2 x 1 )(x 1 y 2 )xCompObjKMfObjInfoNEquation Native _1128505316QFBB 1  2 "x 2 x 1 []ax 2 y 2 +bx 2 x 1 +x 2  1 ax 1 y 2 +bx 1  2 +x 1  1 []=1(x 2 y 2 )(x 1  2 )"(x 2 x 1 )(x 1 y 2 )a((x 2 y 2 )(x 1  2 )"(x 2 x 1 )(x 1 y 2 ))+(x 2  1 )(x 1  2 )"(x 1  1 )(x 2 x 1 )().Ole CompObjPRfObjInfoSEquation Native   FMicrosoft Equation 3.0 DS Equation Equation.39q0|D plim[2a]=a+0x 1  2 +0x 2 x 1  2  1  2 + 2  2 =a_1128513246O^VFBBOle CompObjUWfObjInfoX FMicrosoft Equation 3.0 DS Equation Equation.39qv D 2y  2 ="  1 x 1 +"  2 x 2Equation Native _1128754172[FBBOle CompObjZ\f FMicrosoft Equation 3.0 DS Equation Equation.39qzD 2y  2 FMicrosoft Equation 3.0 DS Equation Equation.39qObjInfo]Equation Native :_1128513373`FBBOle CompObj_afObjInfobEquation Native _1286778688;heFBBh D y 1 =a2y  2 +bx 1 + 1 FMicrosoft Equation 3.0 DS Equation Equation.39q8@>tL 2ay 2 +Ole CompObjdffObjInfogEquation Native \2bx 1 FMicrosoft Equation 3.0 DS Equation Equation.39q8Dt  2a2y  2 +2bx 1Oh+'0_1286778725jFBBOle CompObjikfObjInfolEquation Native `1TableSummaryInformation(oDocumentSummaryInformation88@= D-cAbrI<( *J xZ}l;{>s 8!B+*T\⏊kZEQP VKRE嫘:Ii4 !6Q %R]G.1ݞn!]inggf淿7_ 4}q [0z}72Lz2v/etp]λN4,W+[%}W?6Vw0XB-`=}ֿ0G/r isF!fb;eA1ծZe@g.h[ ձYP"R"V) :Dk;T1'zcYq`MU\vSݺL~=;U,(`F1M)隢72Bst[-  F8u7=[l˸/dxuz^^t!uwڽǫk1xyBemOon3N( v?DOjUoj\;#8_T}ǥ"彡|7u3qW\alsۛnLSf^9~~)=DZ̪eLJ}@:Md}ƽT\exXz*ی g%~;%^j,˗t\UW勴F,kD9ZkZf[y[cy3/j~>-c+0#fGeM@W_2C L;ˎD[k;Xx۪ࢳi=z70,ykQahWʠ gʓ8#H&hݭ)"^77#;8P'Y?&,򚦽|SAfp?,U:a8|w3`;(wea_Z=?R+JSN'7c>|~ۦЊ7snl) D+Ԋh`8{Ŗ3bkſKъԊ F+v >lkEKi ZT./ Z?{vÞy\vO oRliq)@QGSskXA?L٪G;S!^ko%ҙb-=&vaq{BMNym"5:=7߷ĸQS1FQ99Lxᣪ6>K1ZdKV"-8d!C ݚ ^(Fr9(ok rb4d&Q3ht|kT_"oCڥ}GQ}Cr֩ܚ,f:/߿ s3z3wv;ܶ ֦I5.(-&}Sߎ jcٖ[cH1NߜcRkSA[6߿ɭydK7 %g0ta(\؀$P{97 O9<Ab˛k~ 7/oq9cQseY5$]%Xnrҝ3XBQ܁TV`%~c`xVsw'>/YǑ|cx?=t6aYV%q5UY*,w}[^tƇ,sR T_ dTiߋB{RtkRbD[<-\}\97"`wEe|!-x[ҹ[f[z]1;9XܱNAMMsdd8ʐD/GzƟu䮵b,P\~Dd $b  c $A? ?3"`?2={31ggȾw``!{31ggȾw> #xڕKhAgf}mDjQH)hVxR/RAۋ GZ4 REɃ(AOFOT!='EA is)KR4k*0|ef#!.Ok. {cäV>C.J V뱛Qq11.^]fJE#Xjq*#ևw;z_"[pmoQiϡt7]n9=6/5Wa<cیb+l?o09A_U {]S_o\ӿ?>i_=HPDТxaBv2Avy!旟Vzi%['-1k@Dd $Hb  c $A? ?3"`?2EP :Gqf ``!^EP :Gq.@pGH",xڕOAggn+T~HhkY JJ" 6H)AIb"5*Jj'$xμMf}͛V!BԳ|z/y{=bZ(ĨE Xs^G-U($ARb79k̻fCX)~BҬ@j>6GC|t7(gai7'*套y !' ڴ[2O@v  xSo:z t-RzHo:z`;~œ-QAO9|뿴\X\X{\xé빭.*rylR༣o'ry7 p_vL޻ߨ69eq{/D&|q4?BZqXVM2҆f`oW?9ǺKxAϱPTOӵl@+oEd:s*}0 \X}ȮүIa~#"uvލa{C}l=3AqV_kqg"AqO"^kޯJYo&~]Cv*p|# ۯ0ٮ& J:smj^ tMpۮYonME=?)TP(DF"dZc{<~)=TV;RDd b  c $A? ?3"`?2z//5|xxM#``!pz//5|xF`Ho*`\>xڕk@߼vu7nkz$xA"Mne- - ^TnmAr7"6?mzXU5/a+=ClG7):g.) Y19s)םl;Kn'~>Uy)/3Py5?Q4qڕԴ.=SA?/I]eⷯeXy޺kDd Tb  c $A? ?3"`?2EY;{S!&``!Y;{SJ@ XJxcdd``^ @c112BYL%bpuېĺRIio(V-)ă.lJluB.93Qw~Źh~/Zwh^T[! `Z)g?k\C~Ƅߖ!~a L*6'KczZ.~|$N($|G?0MSDd b   c $A ? ?3"`? 2d-kY DW@-``!8-kY DW񄟦 2`\xڝAk@߼dwl즻{P)=HœEbwO]VB+T"z"xW O z$'K:t$/3o&#z(o0UdwqB!JK`dwu1vSZǫ 8 C 9c?n2pqOo*#o?PM a^xjICK'yGFZx)N`N2]P[*d_ Ҽ;eV=Sߵfߪ¿b2<OOp^uCyt4\]T ynk^Ck׹X5|ΆW{5WtNctm=)('r4=gQoDd Hb   c $A ? ?3"`? 2 ZMSqGZa 0``!ZMSqGZa @ `\xڝKP]_i:vh;E*ZpJMElDQZt(])u_,`|Dt1!}}r{aA W{2fY k ZkA[B]ΆB :24!/w AI$BH33nGXz :?1n3FzHaRJj|Z)S[ןUI+'񮌒h&壠sRKSK^iWMYo"=_;lN)RZTWk[&XǤA3^&o}!?#yU5Z>̾Kq}nwa6x!!;ߧ( k*i|` 1jRE 8%pEVTo Dd @Tb   c $A ? ?3"`? 2T-MwZμfW0j3``!(-MwZμfWj  XJxcdd`` @c112BYL%bpu_?:/6:O7hu^+Fq%m搉7j\]\X:w5IL08OB|ϼtOtWe(%s^0=cm6etc-Y=3?1eGFLsSđ'H}ɳ i5 "-)&<}=3\V1Dd Tb   c $A ? ?3"`? 2{H^VC=W8``!OH^VC=  dXJxcdd``~ @c112BYL%bpuePdh,;.Ff =TRDd 4b  c $A? ?3"`?2T"`̣wx:``!pT"`̣w 3 `AOp>xڕ]HQswsg\33]s@(K,AHQz)1( {%0PՇzH|#ߜ93ȃ6@ qC\aNŪǼ.]lU|bgt8]!2!#Xl5"&?A¾#u%y,8ۂo՚ZuzP)Xap^l?J%[3XRG%[/%&$YrWi%ٜToKv( /_rBɎdII'D[j!*٪Ge[_zCT򒭗4)޴͞)2XO)銷CcKLaz--" *7"^g:Q pTNßT$ C"_U?A\QynB,5~B{CsxkBF|%yy I1'aLó >Ox.T(q VCvWAU sE`lcψT͜+l^s{> 7q?P}$$E*ܧhl?A|78ȿ\~uȩ YnX:/ 9nrD|/8)[q|Sۄc$gۏ z!?0!-FOo,IGEg&ǏeDd b  c $A? ?3"`?2FV`K.r)h 4"H?``!V`K.r)h 4J`!Hxcdd``^ @c112BYL%bpub 95_Y͗Пc X?׏'psA72=4?_8?e1rOdu(Qm~8QNp'>ڍf0H <Z>3A1_msC+A[mc}?0  ͹N1#דX'0.ioOk0oB?n̏=ԏ?j8U=~qa_xty|ߪ"j]k?,ܭyt58A=[<1ߢw./y|ևZoZuKzgWpr%:ǟB2; hev xo&;]pEջ7e,Ǚ8 K7G^),_ Dd b  c $A? ?3"`?2&#Xgz1yF``!#Xgz1A`\xڝ+Q=3xsBx$%JI DzedPOYRV66ollPkŸXQ=ΛfL{Νg=31AQ! "˨5/Ɉ2 L8.rO Z'*aNjcQ `k~``ÈI)$5bPd(/v<^ nFf,}MdDd Tb  c $A? ?3"`?2ַ963gM``!ַ963g` XJPxcdd``ved``baV d,FYzP1n:&n! KA?H1Z ǀqC0&dT20ͤ KXB2sSRs=^~*g`0``F\rVy, <;糁&? 2g w%D8ߍP 27)?KW@w$8|R0rp~#+*8|l+Yp@LN LLJ% "CX1Yaa r$RDd |b  c $A? ?3"`?2'@Rex^sgx6P``!p'@Rex^sgJ`` 0>xcdd``Ved``baV d,FYzP1n:6! KA?H1 l @P5< %! `3v ,L ! ~ Ay +V}. __peF\ LYAZ +a`UBo |Fn, #**a|a0? ?3p2Z1@cq' W&00 ~2B  \4=edbR ,.Iex2C D,ĀF`DpDd ,Tb  c $A? ?3"`?2Z=Pa(6T``!.=Pa(â XJxcdd`` @c112BYL%bpucQ `k~``ÈI)$5bPd(/v<^ nFf,}M $0 P \ h t$Simultaneous Equation Regressionjhess Normal.dotjhess8Microsoft Office Word@@꿓@&Ϝ@.ަBT՜.+,0 hp|  uhZ  !Simultaneous Equation Regression Title  FMicrosoft Office Word Document MSWordDocWord.Document.89q@@@ NormalCJ_HaJmH sH tH DA@D Default Paragraph FontRi@R  Table Normal4 l4a (k@(No ListHH "= Balloon TextCJOJQJ^JaJ4 @4 !rFFooter  !.)@. !rF Page Number4@"4 Header  !>@2> (p Footnote TextCJaJ@&@A@ (pFootnote ReferenceH*@Cx/GdUPu 4 1pr s & ' a * B C &?@A'ybc@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@00@000@@0I00@0I00@0I00@0I00@0y00y00y00@0@0y00I0084 pr ybK00I00K00K00K00YI00I00@<@ 5<@ 5 *5BBBE p(P! "&OUXz\chh !#$56789;!Yh":h\pr * > @ %'&:<K_a.BDy{::::::::::::::::::::::*15<>E!!8@0(  B S  ?bcbcbcc87R*  #/oW1!<"=|B!rFJSUFWmY\{`ik m(p)u:} B~M[$@mMO\l',!?; rbyor.{U5!.:'u2D y.S-m6Py}@6-}}} } } pp p@pppp(@ppppp<@p p"p$p&pJUnknownGz Times New Roman5Symbol3& z Arial5& zaTahoma"1hz&u چDz&TZTZ#4d  2QHX ? 2 Simultaneous Equation RegressionjhessjhessCompObjq