EC 2104 Quantitative Methods for Economic Analysis

Problem Set 4 Suggested Solutions

Students will solve selected problems from the following questions during tutorial sessions the week
of February 18, 2019.

1. Find all the first-order partial derivatives of the following functions:

(a) f (x, y) = 5x3 ey ln x

fx0 (x, y) = 5ey (x3 · 1/x + 3x2 ln x) = 5x2 ey (1 + 3 ln x) (Product Rule)

fy0 (x, y) = 5x3 ey ln x (Exponential Rule)

(b) f (x, y) = (ax + by)2 , where a and b are constants.

fx0 (x, y) = 2a(ax + by) (Power Rule)

fy0 (x, y) = 2b(ax + by)

z
(c) f (x, y, z) = xy

For fx0 , we treat y z as a constant A, so we have f (x) = xA . We can apply to power rule,
f 0 (x) = Ax(A−1) , to obtain:
z
fx0 (x, y, z) = y z x(y −1)

For fy0 , we treat x and z as constants A and B. Then we have f (y) = Ag(y) , where g(y) =
y B . Applying the exponential and chain rules, we get that f 0 (y) = Ag(y) · ln A · g 0 (y).
Now, g 0 (y) = By (B−1) , using the power rule. This gives us:
z
fy0 (x, y, z) = xy (ln x) · z · y (z−1)

Finally, for fz0 , we use the same technique as above with fy0 . We have f (z) = Ah(z) ,
where h(z) = B z (using x = A and y = B). We still get that f 0 (z) = Ah(z) · ln A · h0 (z).
Here, h0 (z) = B z · ln B using the exponential rule again. This gives us:
z
fz0 (x, y, z) = xy (ln x) · y z ln y

1
 1
(d) F (K, L) = [aK r + (1 − a)Lr ] r

1 1
0 (K, L) = 1 [aK r + (1 − a)Lr ] r −1 · a · rK (r−1) = aK (r−1) [aK r + (1 − a)Lr ] r −1
FK r
1 1
FL0 (K, L) = 1r [aK r + (1 − a)Lr ] r −1 · (1 − a) · rL(r−1) = (1 − a)L(r−1) [aK r + (1 − a)Lr ] r −1

n
Y
(e) u(x1 , x2 , . . . , xn ) = (xi − γi )βi = (x1 − γ1 )β1 (x2 − γ2 )β2 . . . (xn − γn )βn ,
i=1
where γi and βi are constants for all i.

We perform the derivative on a generic xj , keeping all other x’s constant. In this first
step, we extract the term of the function containing xj . The rest of the function (the
part after the multiplication symbol below) is unaffected:

Y
u0j (x1 , x2 , . . . , xn ) = βj (xj − γj )(βj −1) · (xi − γi )βi
i6=j

This next step is not necessary, but can make the solution easier to read. We wish to
insert the term containing xj back into the part of the function in the multiplication
symbol. Since the new xj term has the power (βj − 1) while the other terms have the
powers βi , we rewrite our solution to obtain:

(xj − γj )βj Y
u0j (x1 , x2 , . . . , xn ) = βj · (xi − γi )βi
xj − γj
i6=j
n
βj Y
= · (xi − γi )βi
xj − γj
i=1

2. Find the Hessian of f (x, y, z) = xy 2 + yz 2

First, we must find the first-order partial derivatives:

fx0 (x, y, z) = y 2

fy0 (x, y, z) = 2xy + z 2

fz0 (x, y, z) = 2yz

Now, we can find all the second-order partial derivatives. I will only find 6 out of the 9 second
derivatives, since Young’s theorem tells us the other cross-partials will be identical. You can
check this yourself.
00
fxx (x, y, z) = 0

2
 00
fxy (x, y, z) = 2y
00
fxz (x, y, z) = 0
00
fyy (x, y, z) = 2x
00
fyz (x, y, z) = 2z
00
fzz (x, y, z) = 2y
The Hessian is:
   
00 (x, y, z) f 00 (x, y, z) f 00 (x, y, z)
fxx 0 2y 0
xy xz
   
   
00
   
f (x, y, z) =  00 00 00  =  2y 2x 2z
 fyx (x, y, z) fyy (x, y, z) fyz (x, y, z)

  
   
   
00 (x, y, z) f 00 (x, y, z) f 00 (x, y, z)
fzx 0 2z 2y
zy zz

3. Question 11.8.3 on page 408 of the textbook. Reproduced here:

Let z = xp1 · · · xpn exp(a1 x1 + · · · + an xn ), where a1 , . . . , an , and p are constants. Find the
partial elasticities of z with respect to x1 , . . . , xn .

First, recall that:

∂z xi
Eli z =
∂xi z
∂z
To find , we keep all other xj , j 6= i, constant. Let’s rewrite z separating xi from all
∂xi
other xj :

z = [xp1 · · · xpi−1 xpi+1 · · · xpn exp(a1 x1 + · · · + ai−1 xi−1 + ai+1 xi+1 + · · · + an xn )] ∗ xpi exp(ai xi )

For simplicity, let’s call the term inside the brackets Q. Now let’s take the partial derivative
with respect to xi and find the elasticity:
∂z
= Q[ai xpi exp(ai xi ) + pxip−1 exp(ai xi )]
∂xi
 
p p
= Qxi exp(ai xi ) ai +
xi
 
p
= z ai +
xi
 
∂z xi p xi
= z ai +
∂xi z xi z
= ai xi + p

So, for any i ∈ 1 . . . n, Eli z = ai xi + p.

3
4. Question 11.3.9 on page 393 of the textbook. Reproduced here:
The following figure shows some level curves for z = f (x, y), together with the line 2x + 3y =
12.

(a) What are the signs of fx0 and fy0 at the points P and Q?

fx0 is the slope of the function when facing the positive x direction. Point P is on
the level curve z = 2. In the positive x direction, the next level curve is z = 3. So, in
that direction, the function is increasing. Therefore, fx0 is positive at P . At point Q, the
function is decreasing in the positive x direction (from z = 4 to z = 3), so fx0 is negative
at Q.
Similarly, fy0 is the slope of the function when facing the positive y direction. At point
P , the next level curve in the positive y direction is z = 1, so fy0 is negative at P . At Q,
the next level curve in the positive y direction is z = 5, so fy0 is positive there.

(b) Find possible solutions of the equations:

i. f (1, y) = 2

We want to find the values of y at which the vertical line x = 1 crosses the level
curve z = 2. However, this does not ever seem to occur. There are no possible
solutions of this equation.

ii. f (x, 2) = 2

We want to find the values of x at which the horizontal line y = 2 crosses the

4
 level curve z = 2. According to the graph of the level curves, this seems to occur
when x = 2 and when x = 6.

(c) What is the largest value of f (x, y) among those (x, y) that satisfy 2x + 3y = 12?

To find the answer to this question, we should look at which level curves the line defined
by 2x + 3y = 12 crosses. Around x = 1.2, the line crosses the z = 1 level curve. As
x increases, the line crosses larger level curves. This indicates that starting at around
x = 1.2, if we increase x, we increase the f (x, y) values that the points on the line
achieve. The line also seems to be tangent to the z = 3 level curve. So, for (x, y) to
the left of the tangent point, f (x, y) < 3. Also, for (x, y) to the right of the tangent
point, f (x, y) < 3 as well. This means that on the line 2x + 3y = 12, f (x, y) achieves
a maximum value of 3. The fact that this occurs when the line is tangent to a level
curve is not a coincidence. Later on, this is how we will maximize functions of multiple
variables given a constraint (in this case 2x + 3y = 12).

5
5. Sample Midterm Question: Which of the following shows some level curves of the function
x2
f (x, y) = y − , with x ≥ 0 and y ≥ 0?
4

y y

(A) (B)

x x

y y

(C) (D)

x x

(E) None of the above.

x2 x2
Solution: (C). A level curve of f (x, y) = y − is of the form y − = c. Solving for y, this
4 4
x2
gives y = + c, which should look like y = x2 shifted and stretched, giving the answer (C).
4