{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

multiequationgmmslides

# multiequationgmmslides - Multiple Equation Linear GMM Eric...

This preview shows pages 1–12. Sign up to view the full content.

Multiple Equation Linear GMM Eric Zivot November 23, 2009 Multiple Equation Linear GMM

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Notation y iM , i = individual; M = equation There are M linear equations, y im = z 0 im (1 × L m ) δ m ( L m × 1) + ε im , m = 1 , . . . , M ; i = 1 , . . . , n Remarks: 1. No a priori assumptions about cross equation error correlation 2. No cross equation parameter restrictions
Giant Regression Representation y 1 n × 1 . . . y M n × 1 = Z 1 n × L 1 . . . Z M n × L M δ 1 L 1 × 1 . . . δ M L M × 1 + ε 1 n × 1 . . . ε M n × 1 or y ¯ nM × 1 = Z ¯ nM × L δ L × 1 + e ¯ nM × 1 L = M X m =1 L m

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Main Issues and Questions: 1. Why not just estimate each equation separately? (a) Joint estimation may improve e ciency (b) Joint estimation is sensitive to misspeci fi cation of individual equations 2. Theory may provide cross equation restrictions (a) Improve e ciency (b) test restrictions
Example: 2 equation wage equation LW i = φ 1 + β 1 S i + γ 1 IQ i + πEXPR i + ε i 1 , L 1 = 4 KWW i = φ 2 + β 2 S i + γ 2 IQ i + ε i 2 , L 2 = 3 z i 1 = (1 , S i , IQ i , EXPR i ) 0 z i 1 = (1 , S i , IQ i ) 0 Note, ε i 1 and ε i 2 may be correlated (eg. due to common omitted variable ability)

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Example: Panel data for wage equation LW 69 i = φ 1 + β 1 S i + γ 1 IQ i + π 1 EXPR i + ε i 1 , LW 80 i = φ 2 + β 2 S i + γ 2 IQ i + π 2 EXPR i + ε i 2 , If all coe cients do not change over time then φ 1 = φ 2 , β 1 = β 2 , γ 1 = γ 2 , π 1 = π 2 ε im = α i + η im α i = unobserved individual fi xed e ff ect
Instruments x im ( K m × 1) = instruments for m th equation E [ x im ε im ] = 0 , m = 1 , . . . , M K = M X m =1 K m orthogonality conditions Note: We are not assuming cross-equation orthogonality conditions. That is, we may have E [ x im ε ik ] 6 = 0 unless x im and x ik have variables in common.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Example: 2 equation wage equation Assume 1. IQ is endogenous in both equations 2. MED is exogenous in both equations E [ MED i ε i 1 ] = 0 , E [ MED i ε i 2 ] = 0 x i 1 = x i 2 = (1 , S i , EXPR i , MED i ) 0
GMM Moment Conditions and Identi fi cation De fi ne δ ( L × 1) = ( δ 0 1 , . . . , δ 0 M ) 0 , L = M X m =1 L m g i ( δ ) K × 1 = g i 1 ( δ 1 ) . . . g iM ( δ M ) = x i 1 ε i 1 . . . x iM ε iM = x i 1 ( y i 1 z 0 i 1 δ 1 ) . . . x iM ( y iM z 0 iM δ M ) Then there are K linear moment equations such that E [ g i ( δ )] = 0 E [ g i ( ˜ δ )] 6 = 0 for ˜ δ 6 = δ

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Now, E [ g i ( δ )] = E [ x i 1 y i 1 ] . . . E [ x iM y iM ] E [ x i 1 z 0 i 1 ] δ 1 . . . E [ x iM z 0 iM ] δ M = E [ x i 1 y i 1 ] . . . E [ x iM y iM ] E [ x i 1 z 0 i 1 ] · · · 0 . . . . . . . . . 0 · · · E [ x iM z 0 iM ] δ 1 . . . δ M = σ xy ( K × 1) Σ xz ( K × L ) δ ( L × 1) Note: The above moment conditions for each equation are the same as the moment conditions we derived for the single equation linear GMM model.
If K m = L m for m = 1 , . . . , M then δ is exactly identi fi ed and δ 1 .

This preview has intentionally blurred sections. Sign up to view the full version.

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

{[ snackBarMessage ]}

### What students are saying

• As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

Kiran Temple University Fox School of Business ‘17, Course Hero Intern

• I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

Dana University of Pennsylvania ‘17, Course Hero Intern

• The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

Jill Tulane University ‘16, Course Hero Intern