Macroeconomics analysis II, EC3102

Tutorial 3 Solution
Question 1

Note:

Given the notation rule in this question, t stands for the present (or now); and so 𝑠𝑠 stands for the
number of periods away from “now”. In other words, we are using the letter 𝑠𝑠 as a counter for the
time index.

At time 𝑡𝑡 + 𝑠𝑠, the budget for the period is:

𝑃𝑃𝑡𝑡+𝑠𝑠 𝑐𝑐𝑡𝑡+𝑠𝑠 + 𝑆𝑆𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠 = ���������������������
������������� 𝑆𝑆𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠−2 + 𝐷𝐷𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠−2 + 𝑌𝑌𝑡𝑡+𝑠𝑠 (1)
outflows inflows

Comment: At the time 𝑡𝑡 + 𝑠𝑠:

- The inflows includes the income for that period, the dividend earned from the asset bought two
periods ago (time 𝑡𝑡 + 𝑠𝑠 − 2) and the nominal value of the assets bought two periods ago (this
nominal value is equal to the price of one asset at time 𝑡𝑡 + 𝑠𝑠, which is 𝑆𝑆𝑡𝑡+𝑠𝑠 , multiplied with the
amount of assets bought two periods ago – that is 𝑎𝑎𝑡𝑡+𝑠𝑠−2 .)

- The outflow includes the consumptions (in nominal values) and the amount of money to buy assets
(𝑎𝑎𝑡𝑡+𝑠𝑠 ) the individual wish to hold in this period.

The problem of the consumer is to maximize lifetime utility from time 𝑡𝑡 onwards subject to an
infinite sequence of flow budget constraints, that is:
∞
𝑚𝑚𝑚𝑚𝑚𝑚
{𝑐𝑐𝑡𝑡+𝑠𝑠 , 𝑎𝑎𝑡𝑡+𝑠𝑠 }∞ � 𝛽𝛽𝛽𝛽(𝑐𝑐𝑡𝑡+𝑠𝑠 )
𝑠𝑠=0
𝑠𝑠=0

subject to: �𝑃𝑃𝑡𝑡+𝑠𝑠 𝑐𝑐𝑡𝑡+𝑠𝑠 + 𝑆𝑆𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠 = ���������������������

������������� 𝑆𝑆𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠−2 + 𝐷𝐷𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠−2 + 𝑌𝑌𝑡𝑡+𝑠𝑠 �
outflows inflows 𝑠𝑠=0

Comment: Our choice variables in this maximization problem are 𝑐𝑐𝑡𝑡+𝑠𝑠 and 𝑎𝑎𝑡𝑡+𝑠𝑠 (s = 1,2, … ∞). That
is, we need to choose the right amount for all of these variables in order to maximize the lifetime
utility of this representative individual. Well, at this point, you might wonder how to solve for all of
them because there are infinity numbers of 𝑐𝑐𝑡𝑡+𝑠𝑠 and 𝑎𝑎𝑡𝑡+𝑠𝑠 . Do not worry, we are just solving for one
set of 𝑐𝑐𝑡𝑡+𝑠𝑠 and 𝑎𝑎𝑡𝑡+𝑠𝑠 and then generalize the pattern for the subsequent period’s 𝑐𝑐 and 𝑎𝑎.

The Lagrangian is:

∞

ℒ(𝑐𝑐𝑡𝑡 , 𝑐𝑐𝑡𝑡+1 , … ; 𝑎𝑎𝑡𝑡 , 𝑎𝑎𝑡𝑡+1 , … ; 𝜆𝜆𝑡𝑡 , 𝜆𝜆𝑡𝑡+1 … ) = � 𝛽𝛽 𝑠𝑠 𝑢𝑢(𝑐𝑐𝑡𝑡+𝑠𝑠 )

𝑠𝑠=0
 ∞

− � 𝜆𝜆𝑡𝑡+𝑠𝑠 𝛽𝛽 𝑠𝑠 {𝑃𝑃𝑡𝑡+𝑠𝑠 𝑐𝑐𝑡𝑡+𝑠𝑠 + 𝑆𝑆𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠 − 𝑆𝑆𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠−2 − 𝐷𝐷𝑡𝑡+𝑠𝑠 𝑎𝑎𝑡𝑡+𝑠𝑠−2 − 𝑌𝑌𝑡𝑡+𝑠𝑠 }
𝑠𝑠=0
(2)

Or expanding it, we have:

ℒ(… ) = 𝑢𝑢(𝑐𝑐𝑡𝑡+0 ) + 𝛽𝛽1 𝑢𝑢(𝑐𝑐𝑡𝑡+1 ) + 𝛽𝛽 2 𝑢𝑢(𝑐𝑐𝑡𝑡+2 ) + ⋯

−𝜆𝜆𝑡𝑡+0 𝛽𝛽 0 {𝑃𝑃𝑡𝑡+0 𝑐𝑐𝑡𝑡+0 + 𝑆𝑆𝑡𝑡+0 𝑎𝑎𝑡𝑡+0 − 𝑆𝑆𝑡𝑡+0 𝑎𝑎𝑡𝑡+𝟎𝟎−2 − 𝐷𝐷𝑡𝑡+0 𝑎𝑎𝑡𝑡+𝟎𝟎−2 − 𝑌𝑌𝑡𝑡+0 }
−𝜆𝜆𝑡𝑡+1 𝛽𝛽1 {𝑃𝑃𝑡𝑡+1 𝑐𝑐𝑡𝑡+1 + 𝑆𝑆𝑡𝑡+1 𝑎𝑎𝑡𝑡+1 − 𝑆𝑆𝑡𝑡+1 𝑎𝑎𝑡𝑡+𝟏𝟏−2 − 𝐷𝐷𝑡𝑡+1 𝑎𝑎𝑡𝑡+𝟏𝟏−2 − 𝑌𝑌𝑡𝑡+1 }
−𝜆𝜆𝑡𝑡+2 𝛽𝛽 2 {𝑃𝑃𝑡𝑡+2 𝑐𝑐𝑡𝑡+2 + 𝑆𝑆𝑡𝑡+2 𝑎𝑎𝑡𝑡+2 − 𝑆𝑆𝑡𝑡+2 𝑎𝑎𝑡𝑡+𝟐𝟐−2 − 𝐷𝐷𝑡𝑡+2 𝑎𝑎𝑡𝑡+𝟐𝟐−2 − 𝑌𝑌𝑡𝑡+2 }
−𝜆𝜆𝑡𝑡+3 𝛽𝛽 3 {𝑃𝑃𝑡𝑡+3 𝑐𝑐𝑡𝑡+3 + 𝑆𝑆𝑡𝑡+3 𝑎𝑎𝑡𝑡+3 − 𝑆𝑆𝑡𝑡+3 𝑎𝑎𝑡𝑡+𝟑𝟑−2 − 𝐷𝐷𝑡𝑡+3 𝑎𝑎𝑡𝑡+𝟑𝟑−2 − 𝑌𝑌𝑡𝑡+3 }
−⋯ (3)

Note:

1. Since I have already specified the ℒ function with all the choice variables in equation (2),
so instead of listing out them out again, I just use the ellipsis - that is,ℒ(… ) .

2. Note carefully these terms. The 2nd ellipsis at the end indicate that the summation
continues forever (since the consumer is assumed to maximize lifetime utility), but the terms
written down are only ones that are important for the problem at hand: the consumer in
period t chooses 𝑐𝑐𝑡𝑡 and 𝑎𝑎𝑡𝑡 and there are no other terms in the Lagrangian (i.e., there are no
other budget constraints) that contain these quantities. Also note the as we move successive
periods into the future, the discount factor 𝛽𝛽 is exponentiated further.

To get the first order conditions (FOCs), differentiate ℒ with respect to any set of 𝑐𝑐𝑡𝑡+𝑠𝑠 and
𝑎𝑎𝑡𝑡+𝑠𝑠 (𝑠𝑠 can be any number) and equate the partial derivatives to 0.
ℒ𝑐𝑐 𝑡𝑡+𝑠𝑠 (… ) = 𝛽𝛽 𝑠𝑠 𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠 ) − 𝜆𝜆𝑡𝑡+𝑠𝑠 𝛽𝛽 𝑠𝑠 𝑃𝑃𝑡𝑡+𝑠𝑠 = 0 (4)

and ℒ𝑎𝑎 𝑡𝑡+𝑠𝑠 (… ) = −𝜆𝜆𝑡𝑡+𝑠𝑠 𝛽𝛽 𝑠𝑠 𝑆𝑆𝑡𝑡+𝑠𝑠 − {−𝜆𝜆𝑡𝑡+𝑠𝑠+2 𝛽𝛽 𝑠𝑠+2 [𝑆𝑆𝑡𝑡+𝑠𝑠+2 + 𝐷𝐷𝑡𝑡+𝑠𝑠+2 ]} = 0 (5)

Mathematical comment: From (4) and (5), we need to get rid of 𝜆𝜆𝑡𝑡+𝑠𝑠 and 𝜆𝜆𝑡𝑡+𝑠𝑠+2 because they
are not what we want. But clearly, you can see that there is no way to get rid of 𝜆𝜆𝑡𝑡+𝑠𝑠+2 . So we need
another equation that involves 𝜆𝜆𝑡𝑡+𝑠𝑠+2 . Observe equation (4), we can try the FOC with respect to
𝑐𝑐𝑡𝑡+𝑠𝑠+2 . Fortunately, there is a pattern. As the matter of fact, you can try to partial differentiate ℒ
with respect to 𝑐𝑐𝑡𝑡+𝑠𝑠+2 and see that the result is just a shifting the time index by 2 units. So we get:

ℒ𝑐𝑐 𝑡𝑡+𝑠𝑠+2 (… ) = 𝛽𝛽 𝑠𝑠+2 𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠+2 ) − 𝜆𝜆𝑡𝑡+𝑠𝑠+2 𝛽𝛽 𝑠𝑠+2 𝑃𝑃𝑡𝑡+𝑠𝑠+2 = 0 (6)

From (5), we get:

 𝜆𝜆𝑡𝑡+𝑠𝑠 𝑆𝑆𝑡𝑡+𝑠𝑠 = 𝜆𝜆𝑡𝑡+𝑠𝑠+2 𝛽𝛽2 {𝑆𝑆𝑡𝑡+𝑠𝑠+2 + 𝐷𝐷𝑡𝑡+𝑠𝑠+2 } (7)
𝜆𝜆𝑡𝑡+𝑠𝑠 𝛽𝛽 2 {𝑆𝑆𝑡𝑡+𝑠𝑠+2 + 𝐷𝐷𝑡𝑡+𝑠𝑠+2 }
=
𝜆𝜆𝑡𝑡+𝑠𝑠+2 𝑆𝑆𝑡𝑡+𝑠𝑠

From (4) by (6), we get:

𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠 ) 2
𝜆𝜆𝑡𝑡+𝑠𝑠 𝛽𝛽 𝑠𝑠 𝑃𝑃𝑡𝑡+𝑠𝑠
= 𝛽𝛽 (8)
𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠+2 ) 𝜆𝜆𝑡𝑡+𝑠𝑠+2 𝛽𝛽 𝑠𝑠+2 𝑃𝑃𝑡𝑡+𝑠𝑠+2

Substituting (7) into (8), we get:

𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠 ) 𝛽𝛽 2 {𝑆𝑆𝑡𝑡+𝑠𝑠+2 + 𝐷𝐷𝑡𝑡+𝑠𝑠+2 } 𝛽𝛽 𝑠𝑠 𝑃𝑃𝑡𝑡+𝑠𝑠

= 𝛽𝛽 2
𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠+2 ) 𝑆𝑆𝑡𝑡+𝑠𝑠 𝛽𝛽 𝑠𝑠+2 𝑃𝑃𝑡𝑡+𝑠𝑠+2
𝛽𝛽 2 {𝑆𝑆𝑡𝑡+𝑠𝑠+2 + 𝐷𝐷𝑡𝑡+𝑠𝑠+2 } 𝑃𝑃𝑡𝑡+𝑠𝑠
= (9)
𝑆𝑆𝑡𝑡+𝑠𝑠 𝑃𝑃𝑡𝑡+𝑠𝑠+2

Expressing 𝑆𝑆𝑡𝑡+𝑠𝑠 in term of the rest, we get:

𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠+2 ) 𝑃𝑃𝑡𝑡+𝑠𝑠

𝑆𝑆𝑡𝑡+𝑠𝑠 = 𝛽𝛽 2 � (𝑆𝑆𝑡𝑡+𝑠𝑠+2 + 𝐷𝐷𝑡𝑡+𝑠𝑠+2 ) � (10)
𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠 ) 𝑃𝑃𝑡𝑡+𝑠𝑠+2

The above formula is a general one for a stock price at any time 𝑡𝑡 + 𝑠𝑠. So the present stock price, or
the stock price at time 𝑡𝑡 is:

𝑢𝑢1 (𝑐𝑐𝑡𝑡+2 ) 𝑃𝑃𝑡𝑡

𝑆𝑆𝑡𝑡 = 𝛽𝛽 2 � (𝑆𝑆𝑡𝑡+2 + 𝐷𝐷𝑡𝑡+2 ) � (11)
(𝑐𝑐
𝑢𝑢1 𝑡𝑡 ) 𝑃𝑃𝑡𝑡+2

Economic intuition: if we rewrite equation (9) for time 𝑡𝑡 (that is, now) and also rearranging the term,
we get:

c.
From (11), we observe that the stock price in period 𝑡𝑡 is affected by period 𝑡𝑡 + 2 marginal utility
𝑢𝑢1 (𝑐𝑐𝑡𝑡+2 ), price level 𝑃𝑃𝑡𝑡+2 , stock price 𝑆𝑆𝑡𝑡+2 , and the dividend 𝐷𝐷𝑡𝑡+2 . This is because, by assumption
stated in the question, stock purchased in period 𝑡𝑡 does not result in any returns till period 𝑡𝑡 + 2. So
the decision made in the stock market will be based on the information in period 𝑡𝑡 + 2 only.
Information from other periods does not affect the decision.

Question 2

a.

Note that, in this question, the dividend comes in the following period, 𝑡𝑡 + 1, similar to the
assumption in the lecture note. With dividend tax, the flow budget constraint for period 𝑡𝑡 (that is,
any period) is:

𝑃𝑃 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡−1 + �1 − 𝜏𝜏𝐷𝐷

𝑡𝑡 𝑐𝑐𝑡𝑡 + 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡 = ���������������������
������� 𝑡𝑡 �𝐷𝐷𝑡𝑡 𝑎𝑎𝑡𝑡−1 + 𝑌𝑌𝑡𝑡 (12)
outflows inflows
Comment: In this question, instead of using the time indexing system of (𝑡𝑡 + 𝑠𝑠) where 𝑠𝑠 is the
counter (for this system of indexing, the present corresponds to the time index 𝑡𝑡), we are using the
time indexing system of 𝑡𝑡 where 𝑡𝑡 is the time index counter. So, in this question, the present
corresponds to the time index 0. You might lament why we have to learn both systems. The reason is
these are two ways that economists use in their journal papers and books for time indexing.

Also, you might wonder since the life of this representative individual starts at time 0 (𝑡𝑡 = 0), then
why do we have 𝑎𝑎−1 . Valid question but we can treat 𝑎𝑎−1 as 0. But leaving it in the above budget
constraint would aid our understanding of the BC better.

b.

The problem of the consumer is:

∞
max
{𝑐𝑐𝑡𝑡 , 𝑎𝑎𝑡𝑡 }∞ � 𝛽𝛽 𝑡𝑡 𝑢𝑢(𝑐𝑐𝑡𝑡 )
𝑡𝑡=0
𝑡𝑡=0
∞

subject to: �𝑃𝑃 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡−1 + (1 − 𝜏𝜏𝑡𝑡𝐷𝐷 )𝐷𝐷𝑡𝑡 𝑎𝑎𝑡𝑡−1 + 𝑌𝑌𝑡𝑡 �

𝑡𝑡 𝑐𝑐𝑡𝑡 + 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡 = ���������������������
������� (13)
outflows inflows 𝑡𝑡=0

Lagrangian, using the summation, is:

∞

ℒ(𝑐𝑐0 , 𝑐𝑐1 , … ; 𝑎𝑎0 , 𝑎𝑎1 , … ; 𝜆𝜆0 , 𝜆𝜆1 … ) = � 𝛽𝛽 𝑡𝑡 𝑢𝑢(𝑐𝑐𝑡𝑡 )

𝑡𝑡=0
∞

− � 𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 {𝑃𝑃𝑡𝑡 𝑐𝑐𝑡𝑡 + 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡 − 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡−1 − (1 − 𝜏𝜏𝑡𝑡𝐷𝐷 )𝐷𝐷𝑡𝑡 𝑎𝑎𝑡𝑡−1 − 𝑌𝑌𝑡𝑡 } (14)
𝑡𝑡=0

The expanded Lagrangian is:

ℒ(… ) = 𝑢𝑢(𝑐𝑐0 ) + 𝛽𝛽1 𝑢𝑢(𝑐𝑐1) + 𝛽𝛽 2 𝑢𝑢(𝑐𝑐2 ) + ⋯

−𝜆𝜆0 𝛽𝛽 0 {𝑃𝑃0 𝑐𝑐0 + 𝑆𝑆0 𝒂𝒂𝟎𝟎 − 𝑆𝑆0 𝑎𝑎−1 − (1 − 𝜏𝜏0𝐷𝐷 )𝐷𝐷0 𝑎𝑎−1 − 𝑌𝑌0 }
−𝜆𝜆1 𝛽𝛽1 {𝑃𝑃1 𝑐𝑐1 + 𝑆𝑆1 𝑎𝑎1 − 𝑆𝑆1 𝒂𝒂𝟎𝟎 − (1 − 𝜏𝜏1𝐷𝐷 )𝐷𝐷1 𝒂𝒂𝟎𝟎 − 𝑌𝑌1 }
−⋯
− 𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 {𝑃𝑃𝑡𝑡 𝑐𝑐𝑡𝑡 + 𝑆𝑆𝑡𝑡 𝒂𝒂𝒕𝒕 − 𝑆𝑆𝑡𝑡 𝑎𝑎𝒕𝒕−𝟏𝟏 − (1 − 𝜏𝜏𝑡𝑡𝐷𝐷 )𝐷𝐷𝑡𝑡 𝑎𝑎𝒕𝒕−𝟏𝟏 − 𝑌𝑌𝑡𝑡 }
𝐷𝐷 )𝐷𝐷
−𝜆𝜆𝑡𝑡+1 𝛽𝛽 𝑡𝑡+1 {𝑃𝑃𝑡𝑡+1 𝑐𝑐𝑡𝑡+1 + 𝑆𝑆𝑡𝑡+1 𝑎𝑎𝑡𝑡+1 − 𝑆𝑆𝑡𝑡+1 𝒂𝒂𝒕𝒕 − (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 𝒂𝒂𝒕𝒕 − 𝑌𝑌𝑡𝑡+1 }

−⋯ (15)
Comment: I coloured the terms so that you can see that the asset for a particular period will appear
again in the budget constraint 1 period ahead. Also note that, BCs for times 𝑡𝑡 and 𝑡𝑡 + 1 and are listed
out so that you can see a pattern in the BCs is repeated for any period in time.

Now let’s get the first order conditions (FOCs) with respect to the variables at any time period - say
period 𝑡𝑡. We have:

ℒ𝑐𝑐 𝑡𝑡 (… ) = 𝛽𝛽 𝑡𝑡 𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) − 𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 𝑃𝑃𝑡𝑡 = 0 (16)

and ℒ𝑎𝑎 𝑡𝑡 (… ) = −𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 𝑆𝑆𝑡𝑡 − �−𝜆𝜆𝑡𝑡+1 𝛽𝛽𝑡𝑡+1 �𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝐷𝐷

𝑡𝑡+1 )𝐷𝐷𝑡𝑡+1 �� = 0 (17)

ℒ𝑐𝑐 𝑡𝑡+1 (… ) = 𝛽𝛽 𝑡𝑡+1 𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) − 𝜆𝜆𝑡𝑡+1 𝛽𝛽 𝑡𝑡+1 𝑃𝑃𝑡𝑡+1 = 0 (18)

From (17), we get:

𝐷𝐷 )𝐷𝐷
𝜆𝜆𝑡𝑡 𝑆𝑆𝑡𝑡 = 𝜆𝜆𝑡𝑡+1 𝛽𝛽[𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ]

𝐷𝐷 )𝐷𝐷
𝜆𝜆𝑡𝑡 𝛽𝛽[𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ]
= (19)
𝜆𝜆𝑡𝑡+1 𝑆𝑆𝑡𝑡

From (16) and (18), we get:

𝛽𝛽𝑡𝑡 𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝜆𝜆𝑡𝑡 𝛽𝛽𝑡𝑡 𝑃𝑃𝑡𝑡

=
𝛽𝛽𝑡𝑡+1 𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝜆𝜆𝑡𝑡+1 𝛽𝛽𝑡𝑡+1 𝑃𝑃𝑡𝑡+1
𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝜆𝜆𝑡𝑡 𝑃𝑃𝑡𝑡
= (20)
𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝜆𝜆𝑡𝑡+1 𝑃𝑃𝑡𝑡+1

Substituting (19) into (20), we get:

𝐷𝐷 )𝐷𝐷
𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝛽𝛽[𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ] 𝑃𝑃𝑡𝑡
=
𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝑆𝑆𝑡𝑡 𝑃𝑃𝑡𝑡+1
𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝐷𝐷 )𝐷𝐷
𝑃𝑃𝑡𝑡
𝑆𝑆𝑡𝑡 = 𝛽𝛽 � [𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ] � (21)
𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝑃𝑃𝑡𝑡+1

If we are interested in how today’s stock is being priced, just replace 𝑡𝑡 with 0 to get:

𝑢𝑢1 (𝑐𝑐1 ) 𝑃𝑃0

𝑆𝑆0 = 𝛽𝛽 � [𝑆𝑆1 + (1 − 𝜏𝜏1𝐷𝐷 )𝐷𝐷1 ] � (22)
𝑢𝑢1 𝑐𝑐0
( ) 𝑃𝑃1

From (21), we can observe that the stock price in any period, 𝑡𝑡, is affected by period 𝑡𝑡 + 1 dividend
𝐷𝐷
tax rate 𝜏𝜏𝑡𝑡+1 . The dividend tax rate determines the future after-tax income flows. All else equal, a
higher dividend tax rate in the future implies a lower stock price today. This is because if people
know that here is higher tax on dividend in one period from now, the stock is less attractive to them
and thus its demand is lower, resulting in a lower stock price today.

Comment on the result:

If we use the time index system with that of Question 1, then equation (21) will become:
 𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠+1 ) 𝑃𝑃𝑡𝑡+𝑠𝑠
𝑆𝑆𝑡𝑡+𝑠𝑠 = 𝛽𝛽 � �𝑆𝑆𝑡𝑡+𝑠𝑠+1 + (1 − 𝝉𝝉𝑫𝑫
𝒕𝒕+𝒔𝒔+𝟏𝟏 )𝐷𝐷𝑡𝑡+𝑠𝑠+1 � � (23)
𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠 ) 𝑃𝑃𝑡𝑡+𝑠𝑠+1

Suppose that Question 1’s assumption on the dividend is that the dividend is realized after 1 period
instead of 2, then equation (11) will be the same with that in the lecture note. That is:

𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠+1 ) 𝑃𝑃𝑡𝑡+𝑠𝑠

𝑆𝑆𝑡𝑡+𝑠𝑠 = 𝛽𝛽 2 � (𝑆𝑆𝑡𝑡+𝑠𝑠+1 + 𝐷𝐷𝑡𝑡+𝑠𝑠+1 ) � (24)
𝑢𝑢1 (𝑐𝑐𝑡𝑡+𝑠𝑠 ) 𝑃𝑃𝑡𝑡+𝑠𝑠+1

If you compare the result in (23) with that of part a (24), what difference do you find beside different
time index system? Do you realize that the difference is in the dividend term? That is, it changes from
𝐷𝐷𝑡𝑡+𝑠𝑠+1 to (𝟏𝟏 − 𝝉𝝉𝑫𝑫
𝒕𝒕+𝒔𝒔+𝟏𝟏 )𝐷𝐷𝑡𝑡+𝑠𝑠+1 . This should not be a surprise to you since now instead of considering
only the dividend of the stock, the individual now considers the after tax dividend (or the effective
dividend) in their asset purchasing decision. So the only place that we need to update in the result is
the dividend.

So actually without going through the Lagrange analysis again, using the intuition discussed above,
we could have written down the new asset pricing equation like that of equation (21). But all is not
wasted, at least we went through the chores of mathematics again as a practice, didn’t we?

c.

If we use the intuition discussed at the end of part b, we can straight away write down the solution
to the asset pricing for this new context. It should be:

𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝐷𝐷 )𝐷𝐷

(1 + 𝜏𝜏𝑡𝑡𝑐𝑐 )𝑃𝑃𝑡𝑡
𝑆𝑆𝑡𝑡 = 𝛽𝛽 � [𝑆𝑆 (1
+ − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ] � (25)
𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝑡𝑡+1 𝑐𝑐
(1 + 𝜏𝜏𝑡𝑡+1 )𝑃𝑃𝑡𝑡+1

Comment: Do you see why? The argument is simple. Since there is a tax on the consumption on each
period. You can think of it as that the consumer is facing a higher price level (or 1 + 𝜏𝜏𝑡𝑡𝑐𝑐 )𝑃𝑃 for any
time 𝑡𝑡). But for the sake of those who might not be so convinced and are interested in the
mathematics behind. We can go through the toils again.

The problem of the consumer is:

∞
max
{𝑐𝑐𝑡𝑡 , 𝑎𝑎𝑡𝑡 }∞ � 𝛽𝛽 𝑡𝑡 𝑢𝑢(𝑐𝑐𝑡𝑡 )
𝑡𝑡=0
𝑡𝑡=0
∞

subject to: �(1 + 𝜏𝜏𝑡𝑡𝑐𝑐 )𝑃𝑃𝑡𝑡 𝑐𝑐𝑡𝑡 + 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡 = ���������������������

������������� 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡−1 + (1 − 𝜏𝜏𝑡𝑡𝐷𝐷 )𝐷𝐷𝑡𝑡 𝑎𝑎𝑡𝑡−1 + 𝑌𝑌𝑡𝑡 � (26)
outflows inflows 𝑡𝑡=0

Lagrangian, using the summation, is:

 ∞

ℒ(𝑐𝑐0 , 𝑐𝑐1 , … ; 𝑎𝑎0 , 𝑎𝑎1 , … ; 𝜆𝜆0 , 𝜆𝜆1 … ) = � 𝛽𝛽 𝑡𝑡 𝑢𝑢(𝑐𝑐𝑡𝑡 )

𝑡𝑡=0
∞

− � 𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 {(1 + 𝜏𝜏𝑡𝑡𝑐𝑐 )𝑃𝑃𝑡𝑡 𝑐𝑐𝑡𝑡 + 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡 − 𝑆𝑆𝑡𝑡 𝑎𝑎𝑡𝑡−1 − (1 − 𝜏𝜏𝑡𝑡𝐷𝐷 )𝐷𝐷𝑡𝑡 𝑎𝑎𝑡𝑡−1 − 𝑌𝑌𝑡𝑡 } (27)
𝑡𝑡=0

The expanded Lagrangian is:

ℒ(… ) = 𝑢𝑢(𝑐𝑐0 ) + 𝛽𝛽1 𝑢𝑢(𝑐𝑐1) + 𝛽𝛽 2 𝑢𝑢(𝑐𝑐2 ) + ⋯

−𝜆𝜆0 𝛽𝛽 0 [(1 + 𝜏𝜏0𝑐𝑐 )𝑃𝑃0 𝑐𝑐0 + 𝑆𝑆0 𝒂𝒂𝟎𝟎 − 𝑆𝑆0 𝑎𝑎−1 − (1 − 𝜏𝜏0𝐷𝐷 )𝐷𝐷0 𝑎𝑎−1 − 𝑌𝑌0 ]
−𝜆𝜆1 𝛽𝛽1 [(1 + 𝜏𝜏1𝑐𝑐 )𝑃𝑃1 𝑐𝑐1 + 𝑆𝑆1 𝑎𝑎1 − 𝑆𝑆1 𝒂𝒂𝟎𝟎 − (1 − 𝜏𝜏1𝐷𝐷 )𝐷𝐷1 𝒂𝒂𝟎𝟎 − 𝑌𝑌1 ]
−⋯
− 𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 [(1 + 𝜏𝜏𝑡𝑡𝑐𝑐 )𝑃𝑃𝑡𝑡 𝑐𝑐𝑡𝑡 + 𝑆𝑆𝑡𝑡 𝒂𝒂𝒕𝒕 − 𝑆𝑆𝑡𝑡 𝑎𝑎𝒕𝒕−𝟏𝟏 − (1 − 𝜏𝜏𝑡𝑡𝐷𝐷 )𝐷𝐷𝑡𝑡 𝑎𝑎𝒕𝒕−𝟏𝟏 − 𝑌𝑌𝑡𝑡 ]
𝑐𝑐 𝐷𝐷 )𝐷𝐷
−𝜆𝜆𝑡𝑡+1 𝛽𝛽 𝑡𝑡+1 [(1 + 𝜏𝜏𝑡𝑡+1 )𝑃𝑃𝑡𝑡+1 𝑐𝑐𝑡𝑡+1 + 𝑆𝑆𝑡𝑡+1 𝑎𝑎𝑡𝑡+1 − 𝑆𝑆𝑡𝑡+1 𝒂𝒂𝒕𝒕 − (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 𝒂𝒂𝒕𝒕 − 𝑌𝑌𝑡𝑡+1 ]

−⋯ (28)

Now let’s get the first order conditions (FOCs) with respect to the variables at any time period - say
period 𝑡𝑡. We have:

ℒ𝑐𝑐 𝑡𝑡 (… ) = 𝛽𝛽 𝑡𝑡 𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) − 𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 (1 + 𝜏𝜏𝑡𝑡𝑐𝑐 )𝑃𝑃𝑡𝑡 = 0 (29)

and ℒ𝑎𝑎 𝑡𝑡 (… ) = −𝜆𝜆𝑡𝑡 𝛽𝛽 𝑡𝑡 𝑆𝑆𝑡𝑡 − �−𝜆𝜆𝑡𝑡+1 𝛽𝛽𝑡𝑡+1 �𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝐷𝐷

𝑡𝑡+1 )𝐷𝐷𝑡𝑡+1 �� = 0 (30)
𝑐𝑐
ℒ𝑐𝑐 𝑡𝑡+1 (… ) = 𝛽𝛽 𝑡𝑡+1 𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) − 𝜆𝜆𝑡𝑡+1 𝛽𝛽 𝑡𝑡+1 (1 + 𝜏𝜏𝑡𝑡+1 )𝑃𝑃𝑡𝑡+1 = 0 (31)

From (30), we get:

𝐷𝐷 )𝐷𝐷
𝜆𝜆𝑡𝑡 𝑆𝑆𝑡𝑡 = 𝜆𝜆𝑡𝑡+1 𝛽𝛽[𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ]

𝐷𝐷 )𝐷𝐷
𝜆𝜆𝑡𝑡 𝛽𝛽[𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ]
= (32)
𝜆𝜆𝑡𝑡+1 𝑆𝑆𝑡𝑡

From (29) and (31), we get:

𝛽𝛽𝑡𝑡 𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝜆𝜆𝑡𝑡 𝛽𝛽𝑡𝑡 (1 + 𝜏𝜏𝑐𝑐𝑡𝑡 )𝑃𝑃𝑡𝑡

=
𝛽𝛽𝑡𝑡+1 𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝜆𝜆𝑡𝑡+1 𝛽𝛽𝑡𝑡+1 �1 + 𝜏𝜏𝑐𝑐𝑡𝑡+1 �𝑃𝑃𝑡𝑡+1
𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝜆𝜆𝑡𝑡 (1 + 𝜏𝜏𝑡𝑡𝑐𝑐 )𝑃𝑃𝑡𝑡
= 𝑐𝑐 )𝑃𝑃 (33)
𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝜆𝜆𝑡𝑡+1 (1 + 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1

Substituting (32) into (33), we get:

 𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) 𝛽𝛽[𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝐷𝐷 )𝐷𝐷
𝑡𝑡+1 ] (1 + 𝜏𝜏𝑐𝑐𝑡𝑡 )𝑃𝑃𝑡𝑡
=
𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝑆𝑆𝑡𝑡 �1 + 𝜏𝜏𝑐𝑐𝑡𝑡+1 �𝑃𝑃𝑡𝑡+1
𝑢𝑢1 (𝑐𝑐𝑡𝑡+1 ) 𝐷𝐷 )𝐷𝐷
(1 + 𝜏𝜏𝑐𝑐𝑡𝑡 )𝑃𝑃𝑡𝑡
𝑆𝑆𝑡𝑡 = 𝛽𝛽 � [𝑆𝑆𝑡𝑡+1 + (1 − 𝜏𝜏𝑡𝑡+1 𝑡𝑡+1 ] � (34)
𝑢𝑢1 (𝑐𝑐𝑡𝑡 ) �1 + 𝜏𝜏𝑐𝑐𝑡𝑡+1 �𝑃𝑃𝑡𝑡+1

If we are interested in how today’s stock is being priced, just replace 𝑡𝑡 with 0 to get:

𝑢𝑢1 (𝑐𝑐1 ) (1 + 𝜏𝜏𝑐𝑐0 )𝑃𝑃0

𝑆𝑆0 = 𝛽𝛽 � [𝑆𝑆1 + (1 − 𝜏𝜏1𝐷𝐷 )𝐷𝐷1 ] � (35)
𝑢𝑢1 (𝑐𝑐0 ) �1 + 𝜏𝜏𝑐𝑐1 �𝑃𝑃1

From (35), all else equal, an one-off increase in consumption tax rate in period 𝑡𝑡, 𝜏𝜏𝑡𝑡𝑐𝑐 , the higher the
stock price in period 𝑡𝑡, 𝑆𝑆𝑡𝑡 . The intuition is that increase in period 𝑡𝑡 consumption tax relative to that
of period 𝑡𝑡 + 1 would make the consumption in period 𝑡𝑡 less preferred than the consumption in
period 𝑡𝑡 + 1. Substitution effect implies that consumers will consume less in period 𝑡𝑡, and more in
period 𝑡𝑡 + 1. Thus, consumers would consume less and save more. The saving in this model setting is
through the investments in stocks - the only asset in this economy). Higher demand for stock would
drive the stock price, 𝑆𝑆𝑡𝑡 , up.