The Identification Problem

Chapter 19
Notations and Definitions

 The variables are of two types

 Endogenous: those determined within the model;
stochastic
 Predetermined: those determined outside the
model; nonstochastic
Divided into two categories:
 Exogenous (current and lag)
 Lagged endogenous

01/06/22 Prepared by Sri Yani Kusumastuti 2

Type of equation

1. Structural, or behavioral, equations: portray the

structure of an economy or the behavior of an
economic agent (e.g. consumer or producer) 
structural coefficients

2. Reduced-form equation: express an endogenous

variable solely in terms of the predetermined
variables and the stochastic disturbances 
reduced-form coefficients

01/06/22 Prepared by Sri Yani Kusumastuti 3

Example:
 Structural equation

Consumption Function: Ct   0  1Yt  ut 0  1  1

Income Identity: Yt  Ct  I t
 Reduced-form equation

Yt   0   1 I t  wt
0 1 ut
where  0  1  wt 
1  1 1  1 1  1

01/06/22 Prepared by Sri Yani Kusumastuti 4

 Ct   2   3 I t  wt
0 1 ut
where  2  3  wt 
1  1 1  1 1  1
 The reduced-form coefficient, 1 and 3, are
known as impact, or short-run, multiplier,
because they measure the immediate impact
on the endogenous variable of a unit change in
the value of the exogenous variable.

01/06/22 Prepared by Sri Yani Kusumastuti 5

The Identification Problem
 Whether numerical estimates of the
parameters of a structural equation can
be obtained from the estimated reduced
form coefficients?
 If this can be done  equation is identified
 Just, or exact identified: if unique numerical
values of the structural parameters can be
obtained
 Over identified: if more than one numerical value
can be obtained for some of the parameters of
structural equations

01/06/22 Prepared by Sri Yani Kusumastuti 6

  If this cannot be done  equation is
unidentified, or underindentified
 The identification problem arises causes
different sets of structural coefficients, that
is, different models, may be compatible
with the same set of data.

01/06/22 Prepared by Sri Yani Kusumastuti 7

Under Identified (1)

 Demand-Supply Models

Demand function Qtd   0  1 Pt  u1t 1  0

Supply function Qts   0  1 Pt  u2t 1  0
Equilibrium Q Qt
d
t
s

 Both equations are unidentified. There is no

unique way to estimate the structural coefficients

01/06/22 Prepared by Sri Yani Kusumastuti 8

 The equilibrium condition, we obtain

 0  1 Pt  u1t   0  1 Pt  u2t
The equilibrium price: Pt   0  vt
0   0 u2t  u1t
where  0  and vt 
1  1 1  1
The equilibrium quantity: Qt  1  wt
1 0   0 1 1u2t  1u1t
where 1  and vt 
1  1 1  1

01/06/22 Prepared by Sri Yani Kusumastuti 9

Just, or Exact Identified (1)
Demand function Qtd   0  1 Pt   2 I t  u1t
1  0,  2  0
Supply function Qts   0  1 Pt  u2t 1  0
Equilibrium Qtd  Qts
 It is the presence of an additional variable in the
demand function that enables us to identified
the supply function. There is no unique way to
estimate the demand function

01/06/22 Prepared by Sri Yani Kusumastuti 10

 The equilibrium condition:
 0  1 Pt   2 I t  u1t   0  1 Pt  u2t
The equilibrium price: Pt   0  1 I t  vt
0   0 2
where  0  1 
1  1 1  1
u2t  u1t
vt 
1  1

01/06/22 Prepared by Sri Yani Kusumastuti 11

 The equilibrium quantity: Qt   2   3 I t  wt
1 0   0 1  2 1
where  2  3 =
1  1 1  1
1u2t  1u1t
wt 
1  1
The structural coefficoents:  0   2  1 0
3
1 
1

01/06/22 Prepared by Sri Yani Kusumastuti 12

 Demand function Qtd   0  1Pt   2 I t  u1t
1  0,  2  0
Supply function Qts   0  1 Pt   2 Pt 1  u2t
1  0,  2  0
Equilibrium Qtd  Qts

 It is the presence of an additional variable in the

demand and supply function that enables us to
identified the both function.

01/06/22 Prepared by Sri Yani Kusumastuti 13

 The equilibrium condition:
 0  1 Pt   2 I t  u1t   0  1 Pt   2 Pt 1  u2t
The equilibrium price: Pt   0  1 I t   2 Pt 1  vt
0  0 2
where  0  1 
1  1 1  1
2 u2t  u1t
2  vt 
1  1 1  1

The equilibrium quantity: Qt   3   4 I t   5 Pt 1  wt

1 0   0 1  2 1
where  3  4 =
1  1 1  1
1 2 1u2t  1u1t
5 = wt 
1  1 1  1

01/06/22 Prepared by Sri Yani Kusumastuti 14

 The structural coefficoents:
4 5
1  1 
1 2
 2 4 1 5
2  5  2   4
1 2
0  ? 0  ?

01/06/22 Prepared by Sri Yani Kusumastuti 15

Over Identified (1)
Demand function Qtd   0  1 Pt   2 I t   3 Rt  u1t
1  0,  2  0
Supply function Qts   0  1 Pt   2 Pt 1  u2t
1  0,  2  0
Equilibrium Qtd  Qts
 It is the presence of an additional two variable
in the demand function, the supply function is
over identified and the demand function is
exact identified

01/06/22 Prepared by Sri Yani Kusumastuti 16

The equilibrium price and quantity:
Pt   0  1 I t   2 Rt   3 Pt 1  vt
Qt   4   5 I t   6 Rt   7 Pt 1  wt
0  0 2 3
where  0  1  2 
1  1 1  1 1  1
2 1 0   0 1  2 1
3  4  5 
1  1 1  1 1  1
 3 1 1 2 1u2 t  1u1t
6  7  wt 
1  1 1  1 1  1
u2t  u1t
vt 
1  1

01/06/22 Prepared by Sri Yani Kusumastuti 17

  The number of equations is greater than the
number of unknowns. As a result, unique
estimation of all the parameters of the model is
not possible.
 From the reduced form coefficients, we can
obtain two estimates of the price coefficients in
the supply function and there is no guarantee
that these values will be identical.

6 5
1  or 1 
2 1

01/06/22 Prepared by Sri Yani Kusumastuti 18

Rules for Identification
 There are two rules
1. The order condition of identifiability
2. The rank condition of identifiability

 The notations:
M = number of endogenous variables in the model
m = number of endogenous variables in a given
equation
K = number of predetermined variables in the model
including the intercept
k = number of predetermined variables in a given
equation

01/06/22 Prepared by Sri Yani Kusumastuti 19

The Order Condition of Identifiability

 DEFINITION
In a model of M simultaneous equations in
order for an equation to be identified, it must
exclude at least M − 1 variables (endogenous
as well as predetermined) appearing in the
model. If it excludes exactly M − 1 variables,
the equation is just identified. If it excludes
more than M − 1 variables, it is overidentified.

01/06/22 Prepared by Sri Yani Kusumastuti 20

 DEFINITION
In a model of M simultaneous equations, in
order for an equation to be identified, the
number of predetermined variables excluded
from the equation must not be less than the
number of endogenous variables included in
that equation less 1, that is,
K − k ≥ m− 1
If K − k = m − 1, the equation is just
identified, but if K − k > m − 1, it is
overidentified.

01/06/22 Prepared by Sri Yani Kusumastuti 21

The Rank Condition of Identifiability

 In a model containing M equations in M

endogenous variables, an equation is
identified if and only if at least one nonzero
determinant of order (M − 1)(M − 1) can be
constructed from the coefficients of the
variables (both endogenous and
predetermined) excluded from that particular
equation but included in the other equations
of the model.

01/06/22 Prepared by Sri Yani Kusumastuti 22

 General principles of identifiability of a structural
equation in a system of M simultaneous equations:

1. If K − k > m − 1 and the rank of the A matrix is M −

1, the equation is overidentified.
2. If K − k = m − 1 and the rank of the matrix A is M −
1, the equation is exactly identified.
3. If K − k ≥ m − 1 and the rank of the matrix A is less
than M − 1, the equation is underidentified.
4. If K − k < m − 1, the structural equation is
unidentified. The rank of the A matrix in this case is
bound to be less than M − 1.

01/06/22 Prepared by Sri Yani Kusumastuti 23

A Test of Simultaneity

 If there is no simultaneous equation, or

simultaneity problem, the OLS estimators
produce consistent and efficient estimators.
 On the other hand, if there is simultaneity,
OLS estimators are not even consistent.
 The methods of twostage least squares
(2SLS) and instrumental variables will give
estimators that are consistent and efficient.

01/06/22 Prepared by Sri Yani Kusumastuti 24

 The simultaneity problem arises because
some of the regressors are endogenous and
are therefore likely to be correlated with the
disturbance, or error, term.
 Therefore, a test of simultaneity is essentially
a test of whether (an endogenous) regressor
is correlated with the error term.
 If it is, the simultaneity problem exists, in
which case alternatives to OLS must be
found; if it is not, we can use OLS. To find out
which is the case in a concrete situation, we
can use Hausman’s specification error test.
01/06/22 Prepared by Sri Yani Kusumastuti 25
Hausman Specification Test
 Demand function:
Qt = α0 + α1Pt + α2 It + α3Rt + u1t
 Supply function:
Qt = β0 + β1Pt + u2t

 where P = price
 Q = quantity
 I = income
 R = wealth
 u’s = error term
01/06/22 Prepared by Sri Yani Kusumastuti 26
 Reduced-form equations:
Pt = 0 + 1 It + 2 Rt + vt
Qt = 3 + 4 It + 5 Rt + wt

01/06/22 Prepared by Sri Yani Kusumastuti 27

 The Hausman test involves the following steps:
Step 1. Regress Pt on It and Rt to obtain ˆvt .
Step 2. Regress Qt on Pˆt and vˆt and perform a t
test on the coefficient of ˆvt. If it is significant, do
not reject the hypothesis of simultaneity;
otherwise, reject it. For efficient estimation,
however, Pindyck and Rubinfeld suggest
regressing Qt on Pt and ˆvt

01/06/22 Prepared by Sri Yani Kusumastuti 28

Tests for Exogeneity

 We noted earlier that it is the researcher’s

responsibility to specify which variables are
endogenous and which are exogenous.
 This will depend on the problem at hand and
the a priori information the researcher has.
 But is it possible to develop a statistical test
of exogeneity, in the manner of Granger’s
causality test.

01/06/22 Prepared by Sri Yani Kusumastuti 29

 Y1i = β0 + β2Y2i + β3Y3i + α1X1i + u1i
If Y2 and Y3 are truly endogenous, we cannot
estimate by OLS.

Y1i = β0 + β2Y2i + β3Y3i + α1X1i + λ2Yˆ2i + λ3Yˆ3i + u1i

Using the F test, we test the hypothesis that λ2 = λ3

= 0. If this hypothesis is rejected, Y2 and Y3 can be
deemed endogenous, but if it is not rejected, they
can be treated as exogenous.

01/06/22 Prepared by Sri Yani Kusumastuti 30