Romer 5e Solutions Manual 06
Romer 5e Solutions Manual 06
Problem 6.1
(a) The IS curve in Figure 6.1 is given by
1
(1) ln Yt a ln Yt 1 rt , where a ≡ – (1/θ)lnβ.
The slope of the IS curve is given by dr/dY for a given Yt+1. Differentiating equation (1) yields
1 1
(2) dYt dr t ,
Yt
and thus the slope is given by
dr t
(3) .
dYt IS Yt
Thus an increase in θ increases the slope in absolute value and makes the IS curve steeper. Intuitively, an
increase in the coefficient of relative risk aversion, θ, means that households are less willing to substitute
consumption across time. Thus any given change in r results in a smaller change in consumption and
therefore output along the IS curve. In addition to the change in slope, an increase in θ reduces the
constant term, a, and thus causes the IS curve to shift down.
The LM curve is given by the combinations of Y and r that satisfy the following equation for a given
level of real money holdings:
1/
Mt 1 rt
(4) Yt / .
Pt rt
Taking the natural log of both sides gives us
M 1 1
(5) ln t ln Yt ln(1 rt ) ln rt .
Pt
The slope of the LM curve is given by dr/dY for a given M/P. From equation (5)
1 1 1 1 1
(6) 0 dYt drt drt .
Yt 1 rt rt
Solving for dr/dY gives us
dr t r (1 rt )
(7) t .
dYt LM Yt
Thus an increase in θ also makes the LM curve steeper.
(b) The IS curve does not depend on χ at all, so it is unchanged. From equation (7) we can see that a
change in χ does not affect the slope of the LM curve. From equation (4) we can see that a decrease in χ
increases the demand for real money balances at a given level of output and the real interest rate. This
implies that the LM curve must shift up. Intuitively, with a fixed real money supply, any given level of
output would now require a higher real interest rate for the demand for money to equal the fixed supply.
Thus the new LM curve associated with the lower value of χ must lie above the old LM curve.
(c) Again, the IS curve does not depend on Γ(•) at all, so it does not change. As in the text, consider a
balanced budget change in Mt/Pt and Ct. Specifically, suppose the household raises Mt/Pt by dm and
lowers Ct by [it/(1 + it)]dm. At the margin, this change must not affect utility. The utility cost of this
change is still U'(Ct)[it/(1 + it)]dm. With our modification to the utility function, the utility benefit is now
given by B Γ'(Mt/Pt)dm. Thus the first-order condition for optimal money holdings is now
6-2 Solutions to Chapter 6
M i
(8) B' t t U (C t ) .
Pt 1 i t
Given the form of Γ(•) and U(•) , and the fact that Ct = Yt this implies
M i
(9) B t t Yt .
Pt 1 it
Dividing both sides of equation (9) by B and then taking both sides of the resulting expression to the
exponent -1/χ, we have
1/
Mt 1 rt
(10) B1 / Yt / ,
Pt rt
where we have also used the fact that with the price level fixed, the real and nominal interest rates are
equivalent. Intuitively, a decrease in B reduces the utility from holding any given quantity of real money
balances. Thus, as we can see from equation (10), it reduces optimal money holdings for any given level
of output and the real interest rate. This implies that the LM curve must shift down. With the lower value
of B, any given level of output must now be associated with a lower real interest rate for real money
demand to remain equal to the fixed real money supply.
Problem 6.2
(a) We need to find the average cost per unit time of conversions plus foregone interest, which we will
denote AC. The cost of conversions in nominal terms is P times C and the number of conversions per
unit time is 1/τ. Average foregone interest per unit time is average money holdings, αYPτ/2 multiplied
by the nominal interest rate, i. Thus,
PC YP
(1) AC i.
2
The first-order condition for τ is
AC PC YP
(2) i0.
2 2
This simplifies to
C Yi
(3) .
2 2
Solving for the optimal choice, τ* yields
12
2C
(4) * .
Yi
Note that 2 AC 2 2 PC 3 0 and so τ* is a minimum.
Problem 6.3
(a) Substituting the consumption function, Ct = a + bYt-1, and the assumption about investment,
It = Kt* – cYt-2, into the equation for output, Yt = Ct + It + Gt, yields
(1) Yt = a + bYt-1 + Kt* – cYt-2 + Gt.
Substituting for the desired capital stock, Kt* = cYt-1 , and for the constant level of government
purchases, Gt = G , yields
(2) Yt = a + bYt-1 + cYt-1 – cYt-2 + G .
Collecting terms in Yt-1 gives us output in period t as a function of Yt-1, Yt-2 and the parameters of the
model:
(3) Yt = a + (b + c)Yt-1 – cYt-2 + G .
(b) With the assumptions of b = 0.9 and c = 0.5, output in period t is given by
(4) Yt = a + 1.4Yt-1 – 0.5Yt-2 + G .
Throughout the following, the change in a variable represents the change from the path that variable
would have taken if G had simply remained constant at G .
In period t,
(5) Yt = a + 1.4Yt-1 – 0.5Yt-2 + G + 1,
and thus the change in output from the path it would have taken is given by
(6) Yt = +1.
In period t + 1, using the fact that equation (4) will hold in all future periods,
(7) Yt+1 = 1.4Yt – 0.5Yt-1 = 1.4(+1) – 0.5(0) = +1.4.
In period t + 2,
(8) Yt+2 = 1.4Yt+1 – 0.5Yt = 1.4(+1.4) – 0.5(+1) = +1.46.
In period t + 3,
(9) Yt+3 = 1.4Yt+2 – 0.5Yt+1 = 1.4(+1.46) – 0.5(+1.4) = +1.344.
With similar calculations, one can show that Yt+4 = +1.15, Yt+5 = 0.938 and so on. Thus output
follows a "hump-shaped" response to the one-time increase in government purchases of one. The
maximum effect is felt two periods after the increase in G and the effect then goes to 0 over time.
6-4 Solutions to Chapter 6
Problem 6.4
Problem 6.5
Suppose the increase in g occurs in time period t and define g gH – gL > 0. In what follows, the
change in a variable refers to the difference between its actual value and the value it would have had in
the absence of the rise in g.
(a) In this case, unemployment does not need to change in order for price inflation to remain constant.
Wage inflation simply rises by g since it is given by
(1) wt t g t .
Since price inflation is given by
(2) t t 1 (u t u) ,
then remains constant and u remains equal to u .
(8) w w
t 1 t (u t 1 u ) ,
we can see that w t 1 will be g higher if u t 1 u . Thus the period after the shock, unemployment must
return to u̅ if priceinflation is to remain constant. The required path of the unemployment rate is shown
in the figure at right. The rise in g causes a one-period drop in u.
t must rise by g for price inflation to be constant. Since wage inflation is given by
(c) As in part (b), w
(9) wt t 1 (u t u ) ,
this means that unemployment must fall in period t. As in part (b), the required change in the
unemployment rate is
1
(10) u t g .
(d) As in parts (b) and (c), w t must rise by g for price inflation to be constant. To see what this
implies for unemployment in period t, we first need to calculate the change in ĝ t due to the change in g.
Since
(14) ĝ t ĝ t 1 (1 )g t ,
we can write
(15) ĝ t (1 )g .
6-6 Solutions to Chapter 6
Similar analysis for period t + 1 would yield the following required change in the unemployment rate:
1
(25) u t 2 [1 (1 ) (1 ) 2 (1 )]g .
And in general, in period t + s, the change in the unemployment rate required for constant price inflation
is given by
1
(26) u t s [1 (1 ) (1 ) 2 (1 ) s (1 )]g .
Allowing s to go to infinity, we can write
(27) 1 (1 ) (1 ) 2 (1 ) 1 [(1 )(1 2 )] .
Since 0 < < 1, the infinite series 1 + ρ + ρ2 + … converges and we can write
1
(28) 1 (1 ) (1 ) 2 (1 ) 1 [(1 ) ] 11 0 .
(1 )
Thus
(29) lim u t s 0 .
s
In this case, if price inflation is to remain constant when g rises, the unemployment rate must fall at time t
and then rise back until it asymptotically approaches its original value.
Solutions to Chapter 6 6-7
Problem 6.6
(a) Substitute the IS equation,
(1) Yt = − rt/θ
into the money-market equilibrium condition,
(2) m p L(r e , Y) ,
to obtain
r
(3) m p L(r e , t ) .
With P P and e = 0, equation (3) simplifies to
r
(4) m p L(r, t ) ,
where p ln(P) . Differentiating both sides of equation (4) with respect to m gives us
dr 1 dr
(5) 1 L r LY ( ) .
dm dm
Solving for dr/dm yields
dr 1
(6) .
dm 1
Lr LY ( )
1
Since L r 0 , L Y 0 , and < 0, dr/dm < 0. Thus an increase in the nominal money supply does
lower the real interest rate.
As shown in part (b), the first term on the right-hand side of equation (11) is negative. The denominator
of the second term on the right-hand side is negative, as shown previously. Since L e 0 and
r
e
d /dm > 0, the numerator is also negative and thus that second term is positive. Thus dr/dm < 0 again,
meaning that an increase in the nominal money supply, m, lowers the real interest rate. Comparing
equations (9) and (11) we can see that dr/dm is larger in absolute value here. Thus, if expected inflation
increases when the nominal money supply rises – as it does here – a given change in m has a larger effect
on the real interest rate. Thus a given change in r requires a smaller change in m than in part (b).
(d) We can substitute the assumptions that dp/dm = 1 and de/dm = 0 into equation (10) to obtain
dr 1 dr
(12) 1 1 L r e LY ( ) .
dm dm
And thus
dr
(13) 0.
dm
With complete and instantaneous price adjustment, a change in the nominal money supply does not affect
the real interest rate.
Problem 6.7
(a) Recall that the AD curve is derived from the IS and MP curves. The IS curve is given by
(1) y(t) = – [i(t) – (t)]/θ,
with y'(•) < 0 so that the IS curve is downward-sloping in (y, r) space. The monetary-policy rule is given
by
(2) r(t) = r(y(t) – y (t),(t)),
which, with the assumption that y (t) = 0, simplifies to
(3) r(t) = r(y(t),(t)),
with the constraint that the nominal interest rate, i(t) = r(t) + (t), cannot be negative. Given this
constraint, we can treat the MP curve as horizontal over the range of output for which the Federal
Reserve's desired value of r would result in a negative value of i; that is, over the range of y for which
r(y(t),(t)) + < 0 or r(y(t),(t)) < – .
horizontal portion of the MP curve shifts up; the real interest rate at which the constraint of a non-
negative nominal interest rate becomes binding is now higher (at a 5 percent rate of inflation, the Fed
cannot achieve a real interest rate below negative 5 percent; at a 2 percent rate of inflation, the Fed
cannot achieve an r below negative 2 percent).
The new MP curve is labeled MP( ~ ) in the figure. As drawn, at the level of output where planned and
actual expenditures are equal given the Federal Reserve's choice of r, the constraint is just binding; that
is, r(~y, ~ 0 or r(~
) ~ y, ~
) ~
. In terms of the figure, this means that the IS curve intersects the kink
in the MP curve. The fall in inflation raises output from y1 to ~y . Thus over the range of inflation rates
exceeding ~ , a drop in inflation increases output along the AD curve as it does in our standard model.
y
(0) ~y
time
~
y (0)
time
y
(0)
t1 time
~
~y
y(0)
t1 time
Solutions to Chapter 6 6-11
y
~
~y
(0)
time
y(0)
time
the real interest rate rises. Thus output must fall. See the figures above.
Problem 6.8
The MP equation is now given by
(1) rt by t u MP
t ,
where
(2) u MP
t MP u MP MP
t 1 e t .
The IS curve no longer contains a shock term and is given by
1
(3) y t E t [ y t 1 ] rt .
Note that equation (5) holds in all future periods. Denoting ϕ ≡ θ/(θ + b), this means we can write
6-12 Solutions to Chapter 6
(6) y t j E t j[ y t j1 ] u MP t j for j = 1,2,3, …
Taking expectations of both sides of (6) as of time t, and using equation (2) and the law of iterated
projections gives us
j
(7) E t [ y t j ] E t [ y t j1 ] MP u MP
t .
We can now iterate equation (5) forward. First, substitute for E t[Y t + 1] to obtain
(8) y t u MP t ( E t [ y t 2 ] MP u MP t ).
Now substitute for E t[Y t + 2] to yield
2
(9) y t u MP t MP u MP
t 2 ( E t [ y t 3 ] 2MP u MP
t ),
and so on. We can therefore rewrite (9) as
(10) y t (1 MP 2 2MP ) u MP n
t lim E t [ y t n ] .
n
Since ϕ < 1, the limit would fail to converge to 0 only if E t[Y t + n] diverged. Since this cannot happen—
agents cannot expect output to diverge—then the limit converges to 0. Thus, equation (10) becomes
/
(11) y t u MP
t .
1 MP
Using the definition of φ ≡ θ/(θ + b) we can write (11) as
1
(12) y t b u MP ,
b MP t
b
or simply
1
(13) y t u MP
t .
b MP
Problem 6.9
(a) Substituting the MP equation given by
(1) r t = by t ,
into the IS equation given by
1
(2) y t E t [ y t 1 ] rt u IS t ,
gives us
b
(3) y t E t [ y t 1 ] y t u IS t .
Equation (3) simplifies to
IS
(4) y t E t [ y t 1 ] ut .
b b
(8) E t [u IS IS
t 1 ] IS u t .
Substituting equation (8) into (7) yields
IS
(9) Au IS
t AISu IS
t ut .
b b
Dividing both sides of equation (9) by u IS
t and collecting the terms in A gives us
(10) 1
IS A .
b b
Equation (10) simplifies to
b IS
(11) A ,
b b
or
(12) A .
b IS
Thus, our solution is
(13) y t u IS
t ,
b IS
which is the same as the solution given by equation (6.35) in the text.
(b) Substituting our conjectures for inflation and output in period t into the AS equation gives us our
first equation:
(14) Cu IS IS
t D t 1 t 1 [Au t B t 1 ] .
Substituting for inflation and output into the new form of the MP equation gives us our second equation:
(15) rt b(Au IS IS
t B t 1 ) c (Cu t D t 1 ) .
Substituting for output in the IS equation yields our third equation:
(16) Au IS IS
t B t 1 E t [Yt 1 ] (1/ )rt u t .
Finally, our conjecture that the solution is of the form
(17) y t Au IS
t B t 1 ,
implies that
(18) E t [ y t 1 ] AE t [u IS
t 1 ] B t .
Substituting equation (8) for E t [u IS
t 1 ] and our conjecture for πt yields our fourth equation:
(19) E t [ y t 1 ] A ISu IS IS
t B(Cu t D t 1 ) .
Problem 6.10
With this change, equation (6.31) in the text becomes
1
(1) y t E t [ y t 1 ] rt u IS
t .
Substituting the MP curve, given by
6-14 Solutions to Chapter 6
(2) rt = byt ,
into equation (1) gives us
1
(3) y t E t [ y t 1 ] by t u IS t .
Collecting the terms in yt yields
b IS
(4)
y t E t [ y t 1 ] u t .
Defining ϕ ≡ [θ/(θ + b)] and multiplying both sides of equation (4) by ϕ gives us
(5) y t E t [ y t 1 ] u IS
t .
Note that equation (5) holds in all future periods so we can write
(6) y t j E t j[ y t j1 ] u IS
t j for j = 1,2,3,…
Since
(7) u IS IS
t ISu t 1 e t ,
IS
where eIS is white noise, taking expectations of both sides of (6) as of time t implies
j IS
(8) E t [ y t j ] E t [ y t j1 ] IS ut .
In equation (8), we have used the law of iterated projections, which states that
E t [ E t j[ y t j1 ]] E t [ y t j1 ] . We can now iterate equation (5) forward. That is, we first express
Et[yt+1] in terms of Et[yt+2] and E t [u IS IS
t 1 ] . We then express Et[yt+2] in terms of Et[yt+3] and E t [u t 2 ] and
so on. Doing this yields
y t u IS IS
t E t [ y t 2 ] ISu t
u IS 2 IS 2 2
2 IS
t ISu t E t [ y t 3 ] ISu t
(9) u IS 2 IS 3 2 2 IS n n
t ISu t ISu t lim E t [ y t n ]
n
u IS n n
t lim E t [ y t n ] .
1 IS n
Since both ϕ and ω are less than one, the second term on the right-hand-side of equation (9) will converge
to 0 unless Et[yt+n] diverges. As discussed in the text, for Et[yt+n] to diverge, agents would have to expect
y to diverge, which cannot happen. Thus assuming lim n n E t [ y t n ] 0 is appropriate.
n
Substituting for ϕ in equation (9) leaves us with an expression analogous to equation (6.37) in the text:
(10) y t b u IS
t ,
1 IS
b
which simplifies to
(11) y t u IS
t .
b IS
From equation (11), we can see that the lower is ω – the less responsive is current demand to
expectations of future output – the smaller is the effect on output of shocks to the IS curve. This is
intuitive because the shocks to the IS curve are serially correlated. If ρIS is positive, then a positive shock
today is expected to raise future output. The less that current demand responds to that increase in
expected future output, the less that output rises today.
Solutions to Chapter 6 6-15
Problem 6.11
(a) Substituting the expression for aggregate demand, y = m – p, into the equation that defines the
optimal price for firms, p* = p + y, yields p* = p + (m – p) or simply
(1) p* = (1 – )p + m.
Substituting the aggregate price level, p = fp*, and m = m' into equation (1) yields
(2) p* = (1 – )fp* + m'.
Solving for p* gives us
(3) p* m .
1 (1 )f
Now substitute equation (3) into the expression for the aggregate price level, p = fp*, to obtain
f
(4) p m .
1 (1 )f
Substituting equation (4) and m = m' into the expression for aggregate demand, y = m – p, yields
f 1 f f f
(5) y m m m ,
1 (1 )f 1 (1 )f
or simply
(1 f )
(6) y m .
1 (1 )f
(b) Substituting equation (3), the expression for a firm's optimal price, into the expression describing the
firm's incentive to adjust its price, Kp*2, yields
2
m
2
(7) Kp * K .
1 (1 )f
We need to plot this incentive to change price as a function of f, the fraction of firms that change their
price. The following derivatives will be useful:
(8)
Kp *2 2K(1 )(m) 2
,
f 1 (1 )f 3
and
(9)
2 Kp *2 6K(1 ) (m)
2 2
.
f 2 1 (1 )f 4
6-16 Solutions to Chapter 6
Thus when > 1, the incentive for a firm to adjust its price is a decreasing function of how many other
firms change their price. See the figure.
Kp*2
(c) In the case of < 1, there can be a
situation where both adjustment by all
firms and adjustment by no firms are Km'2 B
equilibria.
A
Point A is an equilibrium with f = 0.
K2m'2
Consider the situation of a representative
firm at point A. If no one else is changing 1 f
their price, the profits a firm loses by not
changing its price, which are given by
Kp*2 = K2m'2, are less than the menu cost of Z. Thus it is optimal for the representative firm not to
change its price. This is true for all firms and thus no one changing price is an equilibrium.
Point B is also an equilibrium with f = 1. Consider the situation of a representative firm at point B. If
everyone else is changing their price, the profits a firm loses by not changing its price, Kp* 2 = Km'2,
exceed the menu cost of Z. Thus it is optimal for the representative firm to change its price. This is true
for all firms and thus everyone changing price is also an equilibrium.
Solutions to Chapter 6 6-17
Now consider the situation of f = 1 at point B. If everyone else is changing their price, the profits that a
representative firm would lose by not changing its price, Kp*2 = Km'2, are less than the menu cost of Z.
Thus it is optimal for the representative firm not to change its price. This is true for all firms and thus it
cannot be an equilibrium for all firms to change their price.
From this discussion, we can see that the equilibrium in this case is for fraction f EQ of firms to change
their price. If fraction fEQ of firms are changing their price, the profit that a representative firm would
lose by not changing its price is exactly equal to the menu cost, Z. Thus the representative firm is
indifferent and there is no tendency for the economy to move away from this point where fraction f EQ of
firms are changing their price.
Problem 6.12
(a) (y1 , r*(y1 )) is the profit a firm receives at aggregate output level y1, if it charges the profit-
maximizing real price, r*(y1 ). (y1 , r*(y0 )) is the profit a firm receives at aggregate output level y1 if it
continues to charge a real price of r*(y0 ), which was the optimal price to charge when aggregate output
was y0 . Thus G = (y1 , r*(y1 )) – (y1 , r*(y0 )) is the additional profit a firm would receive if, when
aggregate output changes from y0 to y1 , the firm changes its price to its new profit-maximizing level.
This represents, therefore, the firm's incentive to change its price in the face of a change in aggregate real
output.
Using equation (2) to find the second derivative of G with respect to y1 gives us
(4) 2G/y12 = 11(y1, r*(y1)) + 12(y1, r*(y1))[r* '(y1)] +
6-18 Solutions to Chapter 6
Now differentiate both sides of the equation that implicitly defines r*(y0), 2 (y0, r*(y0)) = 0, with respect
to y0 to obtain
(7) 21(y0, r*(y0)) + 22(y0, r*(y0))[r* '(y0)] = 0,
and thus
(8) 21(y0, r*(y0)) = – 22(y0, r*(y0))[r* ' (y0)].
Substituting equation (8) into equation (6) yields
(9) 2 G/y12 |y1=y0 = –222(y0, r*(y0))[r* '(y0)]2 + 22(y0, r*(y0))[r* '(y0)]2
= – 22(y0, r*(y0))[r* '(y0)]2.
Thus, since G and G/y1 evaluated at y1 = y0 are both equal to zero, substituting equation (9) into
equation (1) gives us the second-order Taylor approximation:
(10) G – 22(y0, r*(y0))[r* '(y0)]2[y1 – y0 ]2/2.
(c) The [r* '(y0)]2 component reflects the degree of real rigidity. It tells us how much the firm's profit-
maximizing real price responds to changes in aggregate real output. The 22(y0, r*(y0)) component
reflects insensitivity of the profit function. It tells us the curvature of the profit function and thus the cost
in lost profits from the firm allowing its real price to differ from its profit-maximizing value.
Problem 6.13
(a) Substituting the expression for the nominal wage, w = p, into the aggregate price equation,
p = w + (1 – )l – s, yields p = p + (1 – )l – s. Solving for p yields
(1) p = [(1 – )l – s]/(1 – ).
Substituting the aggregate output equation, y = s + l, and equation (1) for the price level into the
aggregate demand equation, y = m – p, yields
(2) s + l = m – [(1 – )l – s]/(1 – ).
Collecting the terms in l leaves us with
(3) l + [(1 – )l/(1 – )] = m + [s/(1 – )] – s.
Obtaining a common denominator and simplifying gives us
(4) [(1 – ) + (1 – )]l/(1 – ) = m + [1 – (1 – )]s/(1 – ),
or
(5) (1 – )l/(1 – ) = m + [s/(1 – )],
and thus finally, employment is given by
(1 )m s
(6) .
(1 )
Simplifying gives us
(1 )(1 )m (1 )s (1 )s (1 )(1 )m (1 )s
(8) p .
(1 )(1 ) (1 )(1 )
Thus, the aggregate price level is given by
(1 ) m s
(9) p .
(1 )
Substituting equation (6) into the aggregate output equation, y = s + l, and simplifying yields
(1 )m s s s (1 )m s
(10) y s .
(1 ) (1 )
And therefore, output is given by
s (1 )m
(11) y .
(1 )
Finally, to get an expression for the nominal wage, substitute equation (9) into w = p:
(1 ) m s
(12) w .
(1 )
The next step is to see the way in which the degree of indexation affects the responsiveness of
employment to monetary shocks. First, use equation (6) to find how employment varies with m:
(1 )m s
(13) .
m (1 )
Taking the derivative of both sides of equation (13) with respect to gives us
m (1)[1 ] (1 )() ( 1)
(14) 0.
(1 ) 2
(1 ) 2
Thus an increase in the degree of indexation, , reduces the amount that employment will change due to a
given monetary shock.
The next step is to examine the way in which the degree of indexation affects the responsiveness of
employment to supply shocks. First, use equation (6) to find how employment varies with s:
(15) .
s (1 )
Taking the derivative of both sides of equation (15) with respect to gives us
s (1)[1 ] ()() 1
(16) 0.
(1 ) 2 (1 ) 2
Thus an increase in the degree of wage indexation, , increases the amount that employment will change
due to a given supply shock.
where we have used the fact that m and s are independent random variables with variances V m and Vs.
We need to find the value of that minimizes this variance of employment. The first-order condition for
this minimization is
V (1 ) ( 1) 1
(18) 2 Vm 2 Vs 0 .
(1 ) (1 ) 2 (1 ) (1 ) 2
Equation (18) simplifies to
(19) 0 = (1 – )( – 1)Vm + Vs.
Collecting the terms in gives us
(20) [(1 – )Vm + Vs ] = (1 – )Vm .
Thus the optimal degree of wage indexation is
(1 )Vm
(21)
(1 )Vm Vs
Given the result in part (a) – that indexation reduces the impact on employment of monetary shocks but
increases the impact from supply shocks – equation (21) is intuitive. First, if Vs = 0 – so that there are no
supply shocks – the optimal degree of indexation is one. In addition, the larger is the variance of the
supply shocks relative to the variance of the monetary shocks, the lower is the optimal degree of
indexation.
(c) (ii) From equation (28), the variance of employment at firm i is given by
2 2
(1 ) ( i )(1 ) ( i )
(29) Var ( i ) Vm Vs .
(1 ) (1 )
The first-order condition for the value of the degree of wage indexation at firm i, i , that minimizes the
variance of employment at firm i is
Var ( i ) (1 ) ( i )(1 ) ( i )
(30) 2 (1 )Vm 2 Vs 0 .
i (1 ) (1 )
Equation (30) simplifies to
Solutions to Chapter 6 6-21
(31) {(1 – ) – i[(1 – )] + (1 – )}(1 – )Vm = ( + i – )Vs ,
which implies
(32) i2 Vs + i[(1 – )]2 Vm = [(1 – ) + (1 – )](1 – )Vm – ( – )Vs .
Thus i is given by
[(1 ) (1 )](1 )Vm [( )]Vs
(33) i .
2 Vs [ (1 )]2 Vm
(c) (iii) We need to find a value of such that the first-order condition given by equation (31) holds
when i = . That is, we need to find a value of such that if economy-wide indexation is given by , the
representative firm, in order to minimize its employment fluctuations, wishes to choose as well.
Setting i = in equation (31) gives us
(34) (1 – )(1 – )Vm = Vs ,
which implies
(35) [Vs + (1 – )Vm ] = (1 – )Vm .
Thus the Nash-equilibrium value of is
(1 )Vm
(36) EQ .
(1 )Vm Vs
This is exactly the same value of we found in part (b); see equation (21). The value of that minimizes
the variance of aggregate fluctuations in employment is also a Nash equilibrium. Given that other firms
are choosing EQ as their degree of wage indexation, it is optimal for any individual firm to choose EQ as
well.
Problem 6.14
(a) We can use the intuitive reasoning employed to explain equation (11.27) in Chapter 11. Consider an
asset that "pays" – c when the individual climbs a palm tree and pays u when an individual trades and
eats another's coconut. Assume that this asset is being priced by risk-neutral investors with required rate
of return equal to r, the individual's discount rate. Since the expected present value of this asset is the
same as the individual's expected value of lifetime utility, the asset must have price VP while the
individual is looking for palm trees and price VC while the individual is looking for other people with
coconuts.
For the asset to be held, it must provide an expected rate of return of r. That is, its dividends per unit
time plus any expected capital gains or losses per unit time, must equal rV P. When the individual is
looking for palm trees, there are no dividends per unit time. There is a probability b per unit time of a
capital "gain" of (VC – VP) – c; if the individual finds a palm tree and climbs it, the difference in the price
of the asset is VC – VP and the asset "pays" – c at that time. Thus we have
(1) rVP = b(VC – VP – c).
(b) The asset must have price VC while the individual is looking for others with coconuts and must
provide an expected rate of return of r. Thus its dividends per unit time plus any expected capital gains
or losses per unit time, must equal rVC. When the individual is looking for others with coconuts, there
are no dividends per unit time. There is a probability aL per unit time of a capital gain of (VP – VC) + u̅;
if the individual finds someone else with a coconut, trades and eats that coconut, the change in the price
of the asset is (VP – VC) and the asset pays u̅ at that time. Thus we have
(2) rVC = aL(VP – VC + u̅).
(d) For a steady state in which L—the total number of people carrying coconuts—is constant, the flows
out of state C must equal the flows into state C. That is, the number of people finding a trading partner
and eating their coconut per unit time must equal the number of people finding and climbing a tree per
unit time.
The number of people leaving state C per unit time is given by the probability of finding a trading
partner, aL, multiplied by the number of people with coconuts and looking for a trading partner, L. The
number of people entering state C per unit time is given by the probability of finding a tree, b, multiplied
by the number of people looking for a tree, (N - L). For a steady state, these two must be equal. That is,
a steady state requires
(11) (aL)L = b(N – L).
Rearranging equation (11), we have the following quadratic equation in L:
(12) aL2 + bL – bN = 0.
Using the quadratic formula gives us
b b 2 4abN b 9b 2 b
(13) L ,
2a 2a a
where we have used the given condition that aN = 2b. Also note that we can ignore the solution with
L = –2b/a < 0.
(e) For such a steady-state equilibrium, the gain to an individual from climbing a tree, V C – VP, which
represents moving from having the value of being in state P to the value of being in state C, must be
greater than or equal to the cost to the individual of climbing the tree, c. That is, for a steady-state
equilibrium where everyone who finds a palm tree actually climbs it, we require
(14) VC – VP c.
Substituting the steady-state value of L = b/a from equation (13) into the expression for V C – VP given in
equation (10), we have
bc ua (b a ) bc bu
(15) VC VP .
r a ( b a ) b r 2b
Solutions to Chapter 6 6-23
(f) The situation in which no one who finds a tree climbs it is a steady-state equilibrium for any c > 0. If
no one else is climbing a tree when they find one, it is optimal for an individual not to climb a tree when
she finds one. If the individual were to climb a tree and pick a coconut, she would lose c units of utility
with no hope of ever trading with someone else. If she does not climb the tree, she loses no utility. Thus
it is optimal not to climb the tree. The decision process is the same for every individual who comes
across a tree. Thus no one climbing a tree, or L = 0, is a steady-state equilibrium for any c > 0. This
implies that for 0 < c b u̅/(r + b), there is more than one steady-state equilibrium. We have shown two:
L = 0 and L = b/a.
In the situation of multiple equilibria, the one with L = b/a involves higher welfare than the one with
L = 0. We have shown that in part (e), with c bu̅ /(r + b), individuals end up gaining utility each time
they climb a tree. That is why they do it. They know that the utility they will eventually receive by
trading their coconut outweighs the cost of climbing the tree to obtain their coconut. Thus the
equilibrium in which people go through a cycle of searching, climbing, searching, trading and eating,
generates positive utility for the individual. In the equilibrium with L = 0, people never achieve any
positive utility since they never trade and obtain the u̅ units of utility from eating another person's
coconut.
Problem 6.15
(a) The individual's problem is to choose labor supply, Li, to maximize expected utility, conditional on
the realization of Pi . Since Li = Yi, the problem is
(1) max E[(Ci (1 / )Yi ) | Pi ]
Yi
Substituting Ci = PiQi /P and Qi = Li gives us
P Y 1
(2) max E i i Yi Pi .
Yi P
Since only P is uncertain, this can be rewritten as
(3) max EPi P Pi Yi (1 )Yi .
Yi
The first-order condition is given by
(4) E[(Pi /P)|Pi ] – Yi-1 = 0,
or
(5) Yi-1 = E[(Pi /P)|Pi ].
Thus optimal labor supply is given by
6-24 Solutions to Chapter 6
(6) Yi E(Pi P) Pi
1 ( 1)
.
Taking the log of both sides of equation (6) and defining yi lnYi yields
(7) yi = [1/( – 1)]lnE[(Pi /P)|Pi ].
(b) The amount of labor the individual supplies if she follows the certainty-equivalence rule is given by
(in logs)
(8) l i = [1/( – 1)]E[ln(Pi /P)|Pi ].
Since ln(Pi/P) is a concave function of (Pi/P), by Jensen's inequality lnE[(Pi/P)|Pi] > E[ln(Pi/P)|Pi]. Thus
the amount of labor the individual supplies if she follows the certainty-equivalence rule is less than the
optimal amount derived in part (a).
Problem 6.16
(a) Model (i) is given by
(1) yt = a' z t - 1 + bet + vt .
This model says that only the unexpected component of money, et, affects output. Model (ii) is given by
(2) yt = ' z t - 1 + mt + vt .
This model says that all money matters for output.
Substituting the assumption about monetary policy, mt = c' z t - 1 + et, into equation (2) yields
(3) yt = ' z t - 1 + [c' z t - 1 + et ] + vt ,
and collecting terms in z t - 1 gives us
(4) yt = (' + c') z t - 1 + et + vt .
The models given by equations (1) and (4) cannot be distinguished from one another. Given some a' and
b, ' = a' – c' and = b have the same predictions. Intuitively, it is not possible to separate the direct
effect of the z's on output from any possible indirect effect they may have through monetary policy. So it
could be the case that only unexpected money matters and the effect of the z's on output that we observe
is simply their direct effect. However, it could also be the case that the expected component of money
affects output and thus the effect of the z's that we observe consists of both the direct and indirect effects.
Solutions to Chapter 6 6-25
(b) Substituting the new assumption about monetary policy, mt = c' zt-1 + ' wt-1 + et , into model (ii)
yields
(5) yt = ' z t - 1 + [c' z t - 1 + ' wt-1 + et ] + vt ,
or collecting the zt-1 terms gives us
(6) yt = (' + c') z t - 1 + ' wt-1 + et + vt .
In this case, it is possible to distinguish between the two theories. Model (i), only unexpected money
matters, predicts that the coefficients on the w's should be zero. Model (ii), all money matters, does not
predict this. Intuitively, since the w's do not directly affect output, if they are correlated with output it
must be due to their indirect effect through their impact on the money supply.
Problem 6.17
(a) Substitute the aggregate price level, p = qpr + (1 - q)pf , into the expression for the price set by
flexible-price firms, pf = (1 – )p + m, to yield
(1) pf = (1 – )[qpr + (1 – q)pf ] + m.
Solving for pf yields
(2) pf [1 – (1 – )(1 – q)] = (1 – )qpr + m.
Since 1 – (1 – )(1 – q) = q + – q = + (1 – )q, equation (2) can be rewritten as
(3) pf [ + (1 – )q] = (1 – )qpr + m,
and thus finally
(1 )q
(4) p f pr m pr (m p r ) .
(1 )q (1 )q (1 )q
(b) Since rigid-price firms set pr = (1 – )Ep + Em, we need to solve for Ep, the expectation of the
aggregate price level. Taking the expected value of both sides of p = qpr + (1 – q)pf gives us
(5) Ep = qpr + (1 – q)Epf.
Thus we have
(6) pr = (1 – )[qpr + (1 – q)Epf ] + Em.
The rigid-price firms know how the flexible-price firms will set their price. That is, they know that
flexible-price firms will use equation (4) to set their prices. Thus the rational expectation of the price set
by the flexible-price firms is
(7) Epf p r (Em p r ) .
(1 )q
Substituting equation (7) into equation (6) yields
(8) p r (1 )qp r (1 q)p r (Em p r ) Em ,
(1 )q
which implies
(1 )(1 q)
(9) p r (1 )p r Em (Em p r ) .
(1 )q
Defining C [(1 – )(1 – q)]/[ + (1 – )q], we can rewrite equation (9) as
(10) pr = [1 – (1 – ) + C] = ( + C)Em,
or
(11) pr ( + C) = ( + C)Em,
and thus finally
(12) pr = Em.
Rigid-price firms simply set their prices equal to the expected value of the nominal money stock.
6-26 Solutions to Chapter 6
We know that y = m – p. Adding and subtracting Em on the right-hand side of this expression yields
(16) y = Em + (m - Em) – p.
Substituting equation (15) into equation (16) yields
(1 q) (1 )q (1 q)
(17) y (m Em) (m Em) (m Em) ,
(1 )q (1 )q
which simplifies to
(18) y
q
m Em .
(1 )q
(c) (i) From equations (15) and (18), we can see that anticipated changes in m affect only prices.
Specifically, consider the effects of an upward shift in the entire distribution of m, with the realization of
m – Em held fixed. From equation (18) we can see that this will have no effects on real output. In this
case, rigid-price firms get to set their price knowing that m has changed and thus incorporate it into their
price-setting decision.
(c) (ii) Unanticipated changes in m affect real output. That is, a higher value of m given its
distribution—that is, given Em—does raise y as we can see from equation (18). In this case, the rigid-
price firms do not get to observe the higher realization of m and cannot incorporate it into their price-
setting decision and hence the economy does not achieve the flexible-price equilibrium.
In addition, flexible-price firms are reluctant to allow their real prices to change. One can show that
y (1 q)q
(19) m Em .
(1 )q2
Thus a lower value of —a higher degree of "real rigidity"—leads to a higher level of output for any
given positive realization of m – Em. This means that the impact on real output of an unanticipated
increase in aggregate demand is larger the larger is the degree of real rigidity or the more reluctant are
flexible-price firms to allow their real prices to vary.