# Smaller system that determines x and y one way to see

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ation of instrument to w , so estimation more precise (37) Introduction Measurement Error Simultaneous Equations Past Exam Practice Question 2005 Question 5 Past Exam 2005 Question 5 Model: Yt = α1 + α2 Xt + ut (38) Xt = β1 + β2 Yt + vt (39) Zt = γ1 + γ2 Yt + γ3 Xt + γ4 Qt + wt (40) u,v,w have ﬁxed distribution, independent of Q and each other, not autocorrelated Parameters positive & α2 β2 < 1 Introduction Measurement Error Simultaneous Equations 2005 Question 5 5)a)i) Endogenous & Exogenous Variables Determined within system (endogenous): Xt , Yt , Zt Determined outside the system (exogenous): Qt Past Exam Practice Question Introduction Measurement Error Simultaneous Equations Past Exam Practice Question 2005 Question 5 5)a)ii) Simultaneous Equations Bias Equations 1 & 2 form smaller system that determines X and Y. One way to see this: Can determine Y and X reduced forms just from ﬁrst two equations β1 + α1 β2 + vt + β2 ut 1 − α2 β2 α1 + α2 β1 + ut + α2 vt Yt = 1 − α2 β2 Xt = So here we can ignore third equation and get expression for SEB as in textbook! (41) (42) Introduction Measurement Error Simultaneous Equations Past Exam Practice Question 2005 Question 5 5)a)ii) Simultaneous Equations Bias As always, to determine properties of estimator substitute dependent variable (and this is X for this equation!) b2 = = = n ¯ ¯ t =1 (Xt − X )(Yt − Y ) n ¯ )2 t =1 (Yt − Y n ¯ ¯ t =1 (β2 (Yt − Y ) + (vt − v ))(Yt n ¯ )2 t =1 (Yt − Y n ¯ ¯ (vt − v )(Yt − Y ) β2 + t =1 n ¯2 t =1 (Yt − Y ) (43) ¯ − Y) (44) (45) (46) Can’t get closed form expressions for expectations, so look at plim: plim(b2 ) = β2 + n ¯ ¯ t =1 (vt − v )(Yt − Y )) n 1 ¯ plim( n t =1 (Yt − Y )2 ) 1 plim( n (47) Introduction Measurement Error Simultaneous Equations Past Exam Practice Question 2005 Question 5 5)a)ii) Simultaneous Equations Bias plim(b2 ) = β2 + Cov (v , Y ) Var (Y ) (48) As above, use RF for Y to get covariance expression: plim(b2 ) = β2 + To sign “bias”: variances positive α2 > 0 1 − α2 β2 > 0 → “bias” positive 2 α2 σv 2 1 − α2 β2 σY (49) Introduction Measurement Error Simultaneous Equations Past Exam Practice Question 2005 Question 5 5)a)iii) OLS on third equation consistent? Key: is there correlation between regressors and disturbance term w in third equation? Y and X reduced forms depend only on u and v (not w) β1 + α1 β2 + vt + β2 ut 1 − α2 β2 α1 + α2 β1 + ut + α2 vt Yt = 1 − α2 β2 Xt = u and v not correlated with w (by assumption). So Cov (Y , w ) = Cov (X , w ) = 0. Qt exogenous, uncorrelated with disturbance terms. So OLS consistent! (50) (51) Introduction Measurement Error Simultaneous Equations Past Exam Practice Question 2005 Question 5 5)a)iv) Is Qt valid instrument to ﬁt second SF equation? Consider if it satisﬁes conditions for valid instrument: 1 Qt not itself regressor → Yes! 2 cov (Q , u ) = 0 → Yes! 3 Correlated with regressor Yt → No! cov (Q , Y ) = cov (Q...
View Full Document

## This document was uploaded on 03/12/2014 for the course ECON 202 at University of London University of London International Programmes (Distance Learning).

Ask a homework question - tutors are online