Asymptotic Theory Assignment

Advanced Econometrics I

Saira Tatiana Calderón Castillo

Student code - 202010013149
Master in Economics (MSc)
School of Economics and Finance
Universidad EAFIT

April 2020

Lehman (1999) - Second chapter - Section 2

Exercise 2.3
An alternative expression for β̂ given by (2.2.13) is:
Pn
i=1 (Vi − V̄ )(Xi − X̄)
Pn 2
j=1 (Vj − V̄ )

Solution

In the section 2.2 -Aplications- we have example 2.2.3 ”Simple linear regression”, where we
have that:

(2.2.13) Pn
(Vi − V̄ )Xi
Pi=1
n 2
j=1 (Vj − V̄ )

is the standard estimator of β̂ in the simple linear regression model:

(2.2.12)
Xi = α + βVi + Ei

where α and β are unknown regression coefficients, the v’s are known constants, and the E’s
are i.i.d. variables with expectation 0 and variance σ 2 .β have optimality properties when the

1
E’s are normal.

According to exercise 2.3, expression (2.2.13) can be written alternatively as:

Pn
i=1 (Vi − V̄ )(Xi − X̄)
Pn (1)
j=1 (Vj − V̄ )2

So, it means that (2.2.13) and (1) are equivalent. To demonstrate that, it will be proved the
following:

Pn Pn
(Vi − V̄ )Xi i=1 (Vi − V̄ )(Xi − X̄)
Pi=1
n 2
⇐⇒ Pn
j=1 (Vj − V̄ ) j=1 (Vj − V̄ )2

Since both expressions’ denominators are the same, it is sufficient to demonstrate that:

n
X n
X
(Vi − V̄ )Xi ⇐⇒ (Vi − V̄ )(Xi − X̄)
i=1 i=1

First, it will be proved that:

(A)
n
X n
X
(Vi − V̄ )Xi ⇒ (Vi − V̄ )(Xi − X̄)
i=1 i=1

Solution

n
X n
X
(Vi − V̄ )Xi = (Vi Xi − V̄ Xi )
i=1 i=1

According to summation properties, we can break up a summation across a sum:

2
 n
X n
X n
X
(Vi − V̄ )Xi = V i Xi − V̄ Xi
i=1 i=1 i=1

Again, based on summation properties, since V̄ is a constant, we can take it out of the sum:

n
X n
X n
X
(Vi − V̄ )Xi = Vi Xi − V̄ Xi
i=1 i=1 i=1

It is important to remember that the arithmetic mean of variables, such as X and V, are:

(2)
n n
X Xi X Vi
X̄ = ; V̄ =
i=1
n i=1
n
X n Xn
nX̄ = Xi ; nV̄ = Vi
i=1 i=1

So, we can rewrite the second component as nV̄ X̄:

n
X n
X
(Vi − V̄ )Xi = Vi Xi − nV̄ X̄
i=1 i=1

Pn
Again, with (2) it is possible to express nV̄ as i=1 Vi , so:

n
X n
X n
X
(Vi − V̄ )Xi = Vi Xi − X̄Vi
i=1 i=1 i=1

3
Based on summation properties, since X̄ is a constant, we can take it out of the sum:

n
X n
X n
X
(Vi − V̄ )Xi = Vi Xi − X̄ Vi
i=1 i=1 i=1

Considering it is necessary to find common factors to obtain (Vi − V̄ ) and (Xi − X̄), two more
components are going to be added with the sum and substract of nV̄ X̄:

n
X n
X n
X
(Vi − V̄ )Xi = Vi Xi − X̄ Vi + nV̄ X̄ − nV̄ X̄
i=1 i=1 i=1

Based on (2). the fourth component can be written as:

n
X Xi
nV̄

i=1
n


Taking into account that X̄ and V̄ are both constants, therefore their product X̄ V̄ is too, third
component can be written as:

n
X
nV̄ X̄ = V̄ X̄
i=1

then:

4
 n
X n
X n
X n
X n
X
(Vi − V̄ )Xi = Vi Xi − X̄ Vi + V̄ X̄ − V̄ Xi
i=1 i=1 i=1 i=1 i=1
Xn Xn
(Vi − V̄ )Xi = (Vi Xi − X̄Vi + V̄ X̄ − V̄ Xi )
i=1 i=1
Xn Xn
(Vi − V̄ )Xi = (Xi (Vi − V̄ ) − X̄(Vi − V̄ ))
i=1 i=1
n
X n
X
(Vi − V̄ )Xi = (Vi − V̄ )(Xi − X̄)
i=1 i=1

(A) was proved.

Now, it will be proved that:

(B)
n
X n
X
(Vi − V̄ )(Xi − X̄) ⇒ (Vi − V̄ )Xi
i=1 i=1

Solution

n
X n
X
(Vi − V̄ )(Xi − X̄) = (Xi Vi − X̄Vi − V̄ Xi + V̄ X̄)
i=1 i=1
n
X n
X n
X n
X n
X
(Vi − V̄ )(Xi − X̄) = Xi V i − X̄Vi − V̄ Xi + V̄ X̄
i=1 i=1 i=1 i=1 i=1

According to (1), the third term can be written as:

2nd
n n n
X X X Xi
X̄Vi = X̄ Vi = X̄nV̄ = V̄ 
n
i=1 i=1 i=1
n


3rd
n
X n
X
V̄ Xi = V̄ Xi = V̄ nX̄ = nX̄ V̄
i=1 i=1

5
Since X̄ and V̄ are constants, the fourth component can be written as:

n
X
X̄ V̄ = nX̄ V̄
i=1

Then,

n
X n
X n
X
(Vi − V̄ )(Xi − X̄) = Xi Vi − V̄ Xi − 
nX̄
V̄ + 
nV̄X̄
 
i=1 i=1 i=1
Xn Xn Xn
(Vi − V̄ )(Xi − X̄) = Xi Vi − V̄ Xi
i=1 i=1 i=1
n
X n
X
(Vi − V̄ )(Xi − X̄) = (Xi Vi − V̄ Xi )
i=1 i=1
n
X n
X
(Vi − V̄ )(Xi − X̄) = (Vi − V̄ )Xi
i=1 i=1

(B) was proved.

Then, it was proved the equivalence between:

Pn Pn
(Vi − V̄ )Xi i=1 (Vi − V̄ )(Xi − X̄)
Pi=1
n 2
⇐⇒ Pn
j=1 (Vj − V̄ ) j=1 (Vj − V̄ )2

References
[1] Lehmann, E. L., Ed. (1999). Elements of Large-Sample Theory. Springer-Verlag