Scribd
Upload
116 views9 pages

Lecture 10 - Chapter 20 - Simulteneous Equation Methods

This document discusses different approaches to estimating simultaneous equation models: 1. Limited information methods estimate each equation individually, while full information methods estimate all equations simultaneously. Full information maximum likelihood is computationally burdensome. 2. Limited information methods commonly used include ordinary least squares (OLS), indirect least squares (ILS), and two-stage least squares (2SLS). ILS estimates exactly identified equations in three steps, while 2SLS estimates overidentified equations in two steps using instrumental variables. 3. 2SLS generates proxy variables for endogenous regressors that are uncorrelated with the error term. It provides unique solutions and can be used for exactly identified equations as well.

Uploaded by

Rearabetswe Mositsa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
116 views9 pages

Lecture 10 - Chapter 20 - Simulteneous Equation Methods

This document discusses different approaches to estimating simultaneous equation models: 1. Limited information methods estimate each equation individually, while full information methods estimate all equations simultaneously. Full information maximum likelihood is computationally burdensome. 2. Limited information methods commonly used include ordinary least squares (OLS), indirect least squares (ILS), and two-stage least squares (2SLS). ILS estimates exactly identified equations in three steps, while 2SLS estimates overidentified equations in two steps using instrumental variables. 3. 2SLS generates proxy variables for endogenous regressors that are uncorrelated with the error term. It provides unique solutions and can be used for exactly identified equations as well.

Uploaded by

Rearabetswe Mositsa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 9

Approaches to estimation

• Given the following M equations model in M endogenous variables:

=+ + ++ ++ +

=+ ++ +++ +

=+ + ++ +++ +

=+ ++ ++

• Limited information methods – estimation of each equation individually taking into account
the restrictions placed in that equation.

• Full information methods – estimation of all equations simultaneously.


Approaches to estimation
• Full information maximum likelihood method (not common):
o Computational burden
o Non-linear parameter solution
o Specification error

• Limited information method (common):


o Ordinary least squares method (OLS)
o Indirect least squares method (ILS)
o Two-stage least squares method (2SLS)
Approaches to estimation: OLS
Consider the following system of equations:
=+ +++
=++ +++
= + + +++
, , =, =0

• Recursive model
Approaches to estimation: ILS
Indirect Least Squares method applicable for estimating exactly identified
structural equation & the estimates obtained are called ILS estimates.
• ILS requires three steps:
o Obtain the reduced form equations (recall chapter 19).
o Apply OLS to the reduced-form equation individually.
o Obtain the structural parameters of the exactly identified equation, from
the estimated reduced form equation.
ILS Example: Estimating exactly identified equation
Consider the following*: Demand function: +++

Supply function: ++
Recall: Identification problem refers to the possibility of calculating the structural parameters of the simultaneous-equation
model from the reduced form parameters.
 Supply function is exactly identified and demand function is not (hint: Chapter 19 - rules of identification)
• First, obtain the reduced-form equation (RFE)** :

++
++
• Next, use the OLS method to estimate the price and supply equation above.
• Then, obtain the structural parameters of the exactly identified equation:
Supply function ++
= and
*Equilibrium condition: =
**Refer to chapter 19 notes
Approaches to estimation: 2SLS
Two stage least squares method is applicable for estimating consistent structural
parameters for over identified equation.
• 2SLS requires two steps:
Step 1: Regress an endogenous variable on all predetermined variables in the
whole system.
 This involves generating a proxy for the stochastic explanatory variable that
is uncorrelated with the error term i.e. Instrumental variable.
Step 2: Use the predicted values of the endogenous variable (in step 1) to
estimate the over identified equation.
2SLS Example: Estimating over identified equation
Consider the hybrid Keynesian quantity-theory of income determination:
Income function: = + ++++
Money supply function: = +
(&)

Recall: Identification problem refers to the possibility of calculating the structural parameters of the simultaneous-
equation model from the reduced form parameters

Step 1: =+++ ++

=+ ( and are uncorrelated)


Step 2:
=+
=++
= ++
where =
Features of the 2SLS method
 Simple application
 Unlike ILS, 2SLS provides unique solutions to each coefficient of the
structural equation
 Can be used to also estimate an exactly identified equation
 Standard errors can be reported
 If is high in the reduced-form regression (stage 1), the OLS and 2SLS
estimates will be close
Test for simultaneity:
Simultaneity problem arise because some of the explanatory variables are endogenous
and can possibly be correlated with the error term

Hausman specification test:


[Variable of interest (income) is exogenous]
[Variable of interest (income) is endogenous]

Step 1:
Regress the endogenous variable (appearing as explanatory in the equation you are
estimating) on predetermined variables in the whole system and obtain the residual.

Step 2:
Regress the equation of interest on the endogenous variable in step 1 and the residual
obtained. Perform a t-test on the latter.

You might also like