Econ 202a

Spring 2000

Suggested Final Exam Solutions

1

Rational and Near Rational Wage Setting

Firms
1M
n p


w
R
w

Each firm faces a commodity demand of q d =

 
p
p

, where > 1. Labor

productivity is given by T (w) = A + B

+ Cu. So, the production
function is q s = T (w)L and the cost function must take the form C(q s) =
w
q s . Hence, the profit function can be written as
T (w)
i =

pi qis

C(qis)


= pi

wi
T (wi)


qid

(1)

for a monopolist. There are two types of firms, rational and near-rational
ones. Label them i = r , nr , respectively.

1.1

Price choice of firm i

Firm i chooses it price by maximizing (1) with respect to pi . The first order
condition for this problem is
 d

wi
qi
d
qi pi
= 0,
T (wi) pi
which is a sufficient condition since revenues are concave in pi . Rearranging
yields
pi =

1.2

wi
.
1 T (wi )

(2)

Wage choice of firm i

Similarly, maximizing (1) with respect to wi yields the first order condition


T (wi) T 0(wi )wi

=0
T (wi)2
1

so that, in optimum, the Solow condition

wi =

T (wi)
T 0(wi )

(3)

must be satisfied. By the definition of T (), this implies that

 
A + B wwRi
+ Cu
 i 
w=
.
1
B wwRi
w
i

Equivalently,

wi =

1.3

A Cu
(1 )B

 1

wiR .

(4)

Relative profits

Given the facts that wrR = wR = w1 (1 + e) for the rational firms and
R
= w1 for the near rational firms, the two wage choices are equal for
wnr
= e = 0:
1

A Cu R
R
R
wr = wnr =
w
if = e = 0.
(1 )B
Hence, the productivity levels must be equal, too: T (wr ) = T (wnr ) for =
e = 0. Then, however, price choices must be equal: pr = pnr . Therefore,
we can conclude that profits r and nr must be identical if = e = 0 and
relative profits (the ratio of profits) must be equal to one: r /nr = 1.
Suppose these relative profits could be expressed as a function of inflation
. Then, by the arguments above, at zero inflation = e = 0 the optimal
choice of pi and wi are the same for both firms. Therefore, the change in
profits is the same by the envelope theorem so that the ratio of profits remains
at one after the change. This means, however, that the derivative of the ratio
r /nr with respect to must be zero.1 This argument suffices to answer
the question.
1

In general,

r
nr


=

r
nr

nr

r
=0
2
(nr )

for r = nr and r / = nr /.

The same results could, of course, also be derived explicitly. This was not
required. For completeness, relative profits could be derived by noting that


wi
i =
pi
qd
T (wi) i


 

wi
wi
1 M pi
=

1 T (wi) T (wi) n p
p


 1 
1
wi
M 1
1
1
=
,
1

n p
T (wi)
and hence
r
nr

"
=

wr
T (wr )
wnr
T (wnr )

#1
=

wrR
R
w
nr

A+B(wr /wR ) +Cu

A+B(wnr /wR ) +Cu

by (4) and the fact that true productivity is given by T (wi) = A + B wi /w R +
Cu for both firms (where w R = w 1 (1 + e )). Setting = e = 0 immediately
yields r /nr = 1. Taking the derivative with respect to and evaluating the
resulting term at = e = 0 yields zero.

1.4

Short-run wage Phillips curve

If we suppose that denotes the share of near-rational firms, we can obtain

the wage level at period 0 as
w0 = wnr + (1 )wr

1
A Cu
=
[w 1 + (1 )w 1 (1 + e )] .
(1 )B

(5)

Subtracting w 1 /w1 from both sides of (5) yields

w0 w1
=
w1

1.5

A Cu
(1 )B

 1

[1 + (1 ) e ] 1.

(6)

Long-run Phillips curve

Since = e = 0 and = (w 0 w1 )/w1 in the long-run, (6) implies

1 + LR = a [1 + (1 )LR ]
3

(7)

Figure 1: Phase Diagram for Tobins q

for a [(A Cu)/(1 )B] . Thus,

LR =

1a
.
1 a(1 )

(8)

Temporary Technological Progress and Tobins q

A temporary rise in the technology parameter A shifts the qt = 0-schedule

temporarily to the East. Thus, between time t0, when productivity unexpectedly rises, and time T > t0, when it expectedly falls back to its initial level,
the new dynamic system with qt = 0 and Kt = 0 temporarily governs
the dynamics of the capital stock and Tobins q.
The capital stock Kt0 cannot adjust immediately and thus remains at its
initial steady-state level for one period. However, the shadow price of capital,
4

Tobins q, has to jump to a new level which is such that the subsequent levels
of both qs and Ks (s t0 ) will be back at a point on the initial saddle path.
So, at t0 Tobins q jumps discretely to point 2 in figure 1 and it will reach point
3 exactly at date T . Between time t0 and T , both variables qs and Ks must
follow the dynamics of the shifted system which take them first to the south
east and then (once the Kt = 0-schedule is surpassed) to the south west
of the system. Note that the shadow price must not change discontinuously
(jump) at any time after t0, otherwise the capital stock cannot have been
chosen optimally. Between time t0 and T , the capital stock first increases
and then falls while Tobins q keeps falling until T . Once time T is reached,
the system is again governed by the old qt = 0 and Kt = 0-schedules
and the capital stock continues to fall smoothly while the shadow price of
capital qs now continuously rises.

3
3.1

Central Bank with Superior Knowledge

Linear optimal monetary rule

Given aggregate supply

yt = (pt Et1 [pt ]) + ut ,

(9)

mdt = pt + vt ,

(10)

yt = mst vt Et1 [mst vt ] + ut

= mst vt Et1 [mst ] + ut1 + t .

(11)

and aggregate demand

output can be written as

The second step follows from the assumptions on the stochastic processes.
Since the conditional expectations over the linear monetary rule mst = a +
but1 + ct + dvt1 are Et1 [mst ] = a + but1 + d vt1 , it follows that
mst Et1 [mst ] = ct .
Hence, output can be written as
yt = (c + 1)t vt + ut1 .
5

(12)

For c = 1, the disturbance of aggregate supply is completely offset by

the monetary rule so that the variance of output will be minimized. The
parameters a, b, d are irrelevant since their impact is entirely anticipated
by the private sector. This arguments can be made formal by minimizing
V(yt) = E[(ct vt + ut1 + t )2] with respect to a, b, c, d (a, b, d dont
enter the minimization problem).

3.2

Non-linear optimal monetary rule

Allowing for higher order terms such as in a rule like mst = a + but1 +
ct + dvt1 + b0(ut1 )2 + c0(t )2 + d0 (vt1 )2 will result in minimizing V(yt ) =
E[(ct + c0 (t)2 vt + ut1 + t)2 ]. Now neither a, b, d nor a0 , b0, d0 enter,
and setting c = 1 is still optimal. Since the noise t is completely offset by
the choice of c = 1 already, anything else but c0 = 0 would add noise back
in. The same is true for any higher order polynomial.

3.3

Stochastic process of ut

The process
t1 + t
ut = ut1 +

(13)

is an ARMA(1,1) process. Using lag operators, (13) can be rewritten as

1 L
ut = t
1 + L

(14)

or
(1 L)(1 L + 2L2 3 L3 + 4 L4 . . . ) ut = t
(1 L L + L2 + 2 L2 2 L3 3 L3 + . . . ) ut = t
(1 ( + )L + ( + )L2 ( + )2 L3 + . . . ) ut = t.

(15)
(16)

Without going through any further derivations, the first step (15) immediately shows that we found an AR() process with
ut =

ks uts + t

s=1

for some ks . The (not required) further steps show that

ut = ( + )

()s1 uts + t

s=1

in fact.

3.4

New optimal monetary rule

The optimal monetary rule does not change at all. Both the Central Bank
and the Private Sector know all (infinitely many) past realizations of uts
(s = 0, 1, 2, . . . ). So it is still the case that the Private Sector rationally
foresees that the expected deviation of the money supply from its mean is
equal to the unknown part of the money rule, mst Et1 [mst ] = ct + c0 (t )2.
Thus, including any past realizations of uts or vts is entirely irrelevant. It
is still optimal for the central bank to set c = 1 and c0 = 0.

4
4.1

Microfoundations of Money Demand

Interpretation of budget constraint

The individual can choose to allocate her current income Yt to three different
uses: to savings At+1, or to money holding Mt , or to consumption Ct . The
more money Mt she chooses to put aside today (at t), the more of todays
t
income Yt becomes useful. In the limit ( M
), income is fully useful.
Pt
In optimum, the individual will choose some intermediate amount of money
holdings.

4.2

Choice variables

The individual can choose Ct , At+1 , and Mt . Since she is restricted by the
budget constraint, choosing any two of the three implies her choice of the
third. Thus, two intertemporal first-order conditions (Euler equations) will
be enough to pin the optimal consumption, savings and money holdings paths
down.

4.3

First-order conditions

Intertemporal utility Ut =

s=t

st u(Cs )2 can be rewritten as

Ut = u(Ct ) + Ut+1 = u(Ct) + u(Ct+1) + 2Ut+2 .

(17)

t
+ g( M
)Yt
Then, plugging the consumption choice Ct = (1 + r)At + MPt1
Pt
t
Mt
At+1 Pt into (17) for t and t + 1, and maximizing (17) with respect to At+1
and Mt yields the two first-order conditions

and

u0(Ct ) + (1 + r)u0 (Ct+1) = 0

(18)


 

1 0
Mt
1 0
0
u (Ct+1 ) = 0.
u (Ct ) 1 g
Yt +
Pt
Pt
Pt+1

(19)

Using (18) in (19) yields


1g

4.4

Mt
Pt


Yt =

1 Pt
.
1 + r Pt+1

(20)

Money demand under specific functional form for

g()

t
Given (20) and Ys = Y , and using g( M
) = 1k e
Pt
we find

kY e

M
t

Pt

=1

M
t

Pt

t
so that g 0 ( M
) = ke
Pt

M
t

Pt

1 Pt
.
1 + r Pt+1

Observe that the nominal interest rate must be related to the real interest
rate through 1 + it+1 = (1 + r) PPt+1
. Then, taking logs of both sides yields
t
Mt
it+1
= ln kY ln
.
Pt
1 + it+1

(21)

This function is increasing in income and decreasing in the nominal interest

rate, just as any LM -function should be.

P
ts
This relationship is correct for < 1. If, as in the question, Ut =
u(Cs ), we
s=t
must require >
1.
All
the
following
derivations
hold
for
the
more
common
case
where
P
< 1 and Ut = s=t st u(Cs ).
2