Title: Approximate quadratic-linear optimization problem
1Approximate quadratic-linear optimization problem
- Based on
- Pierpaolo Benigno and Michael Woodford
2The Quadratic Approximation to the Utility
Function
3The first-order condition
4The second-order approximation to the utility
function
5The second-order approximation to the constraint
6- Substitute the second-order approximation to the
constraint into the linear term of the
second-order approximation to the utility
function, using the FOC, yields a quadratic
objective function
7The approximate optimization problem
Subject to
8Which is supposed to be(?) a first order
approximation of
9A Linear-Quadratic Approximate Problem
- Begin by computing a Taylor-series approximation
to the welfare measure, expanding around the
steady state. As a second-order (logarithmic)
approximation, BW get
10The Quadratic Approximation to the Utility
Function
11The first-order condition
12The second-order approximation to the utility
function
13The second-order approximation to the constraint
14Approximate optimization
- Substitute the second-order approximation to the
constraint into the linear term of the
second-order approximation to the linear term of
the second-order approximation of the utility
function, using the first-order conditions,
yields a quadratic objective function. - The approximate optimization is to maximize the
quadratic objective function, subject to the
first-order approximation of the constraint. The
first-order condition is equal to the first order
approximation of the FOC of the original problem.
15The Micro-based Neo-Keynesian Quadratic-linear
problem
- Based on
- Pierpaolo Benigno and Michael Woodford
16The Micro-based Quadratic Loss Function
17Welfare measure expressed as a function of
equilibrium production
Demand of differentiated product is a function
of relative prices
18The Deterministic (distorted) Steady State
Maximize with respect to
Subject to constraints on
19- BW show that an alternative way of dealing with
this problem is to use the a second-order
approximation to the aggregate supply relation to
eliminate the linear terms in the quadratic
welfare function.
20A Linear-Quadratic Approximate Problem
- Begin by computing a Taylor-series approximation
to the welfare measure, expanding around the
steady state. As a second-order (logarithmic)
approximation, BW get
21- There is a non-zero linear term in the
approximate welfare measure, unless - As in the case of no price distortions in the
steady state (subsidies to producers that negate
the monopolistic power). This means that we
cannot expect to evaluate this expression to the
second order using only the approximate solution
for the path of aggregate output that is accurate
only to the first order. Thus we cannot determine
optimal policy, even up to first order, using
this approximate objective together with the
approximations to the structural equations that
are accurate only to first order.
22Welfare measure expressed as a function of
equilibrium production
Demand of differentiated product is a function
of relative prices
23The Micro-based Quadratic Loss Function of
Benigno and Woodford
24- There is a non-zero linear term in the
approximate welfare measure, unless - As in the case of no price distortions in the
steady state (subsidies to producers that negate
the monopolistic power). This means that we
cannot expect to evaluate this expression to the
second order using only the approximate solution
for the path of aggregate output that is accurate
only to the first order. Thus we cannot determine
optimal policy, even up to first order, using
this approximate objective together with the
approximations to the structural equations that
are accurate only to first order.
25The Deterministic (distorted) Steady State
Maximize with respect to
Subject to constraints on
26- BW show that an alternative way of dealing with
this problem is to use the a second-order
approximation to the aggregate supply relation to
eliminate the linear terms in the quadratic
welfare function.
27MICROFOUNDED CAGAN-SARGENT PRICE LEVEL
DETERMINATION UNDER MONETARY TARGETING
28MICROFOUNDED CAGAN-SARGENT PRICE LEVEL
DETERMINATION UNDER MONETARY TARGETING
FLEX-PRICE, COMPLETE-MARKETS MODEL
29Complete Markets
Value of portfolio with payoff D
price kernel
30Interest coefficient for riskless asset
Riskless Portfolio
31Budget Constraint
Where T is the transfer payments based on
the seignorage profits of the central bank,
distributed in a lump sum to the representative
consumer
32No Ponzi Games
For all states in t1
For all t, to prevent infinite c
The equivalent terminal condition
33Lagrangian
34Transversality condition
Flow budget constraint
35Market Equilibrium
Market solution for the transfers T
36Monetary Targeting BC chooses a path for M
Fiscal policy assumed to be
Equilibrium is
S.t. Euler-intertemporal condition condition FOC-i
tratemporal condition TVC Constraint
For given
37We study equilibrium around a zero-shock steady
state
38Derive the LM Curve
From the FOC
At the steady state
39Separable utility
Define
The hat variables are proportional deviations
from the steady state variables.
40Similar to Cagans semi-elasticity of money
demand
41We log-linearize around
zero inflation
define
Log-linearize the Euler Equation and transform
it to a Fisher equation
Elasticity of intertemporal substitution
g is the twist in MRS between m and c
42Add the identity
We look for solution given exogenous shocks
43Solution of the system
This is a linear first-order stochastic
difference equation ,where,
Exogenous disturbance (composite of all shocks)
44given
There exists a forward solution
From which we can get a unique equilibrium value
for the price level
This is similar to the Cagan-Sargent-wallace
formula for the price level, but with the
exception that the Lucas Critique is taken care
of and it allows welfare analysis.
45I. Interest Rate Targeting based on exogenous
shocks
Choose the path for i specify fiscal policy
which targets D
Total end of period public sector liabilities.
Monetary policy affects the breakdown of D
between M and B
No multi-period bonds
Beginning of period value of outsranding bonds
End of period, one-period risk-less bonds
46Steady state (around
)
fix
47PRICE LEVEL IS INDETERMINATE
Real balances are unique
Future expected inflation is unique
Is unique
But, neither
Can uniquely be determined!
48To see the indeterminancy, let denote
solution value
v is a shock, uncorrelated with (sunspot), the
new triple is also a solution, thus
Price level is indeterminate under the interest
rule!
49II. Wicksellian Rules interest rate is a
function of endogenous variables (feedback rule)
Vcontrol error of CB
Fiscal Policy
Exogenous
Endogenous
50Steady State
Log-linearize
51We can find two processes
Add the identity
521), 2) and 3) yield
P is not correlated to the path of M money
demand shocks affect M, but do not affect P the
LM is not used in the derivation of the solution
to P.
53FEATURES
- Forward looking
- Price is not a function of i rather , a function
of the feedback rule and the target - suppose
54Additionally
Price level instability can be reduced by raising
, an automatic response.
55Note, also that
, reduces the need for accurate observation of
, almost complete peg of interest rate
56The path of the money supply
By using LM, we can still express
But we must examine existence of a
well-defined demand for money. Theres possibly
liquidity trap
57III. TAYLOR (feedback) RULE
Assume
58Taylor principle
Is predetermined
59Transitory fluctuations in
Create transitory fluctuations in
Permanent shifts in the price level P.
60Optimizing models with nominal rigidities
61(No Transcript)
62(No Transcript)
63First Order Conditions
64Firms Optimization
Nominal
Real
65Natural Level of Output
66Log-linearization of real mc
Partial-equilibrium relationship?
67where
Elasticity of marginal product of labor wrt output
Elasticity of wage demands, wrt to output
holding marginal utility of income constant
68ONE-PERIOD NOMINAL RIDIGITY
Same as before, except for
Y need not be equal to the natural y
69A Neo-Wicksellian Framework
THE IS
Ct consumption aggregate
gross rate of increase in the Dixit-Stiglitz
price index Pt
70Equilibrium condition
A log-linear approximation around a deterministic
steady state yields the IS schedule
gcrowding out term due to fiscal shock
71Effect on fiscal shock on C
Equivalent to the fiscal shock
72New Keynesian Phillips Curve
Deviation of natural output due to supply shock
Demand determined output deviations
Taylor Rule
Inflation target
73Output gap
3-EQUATION EQUILIBRIUM SYSTEM
Proportion of firm that prefix prices
IS-curve involves an exogenous disturbance term
74INTEREST RULE AND PRICE STABILITY
THE NATURAL RATE OF INTEREST
75Percentage deviation of the natural rate of
interest from its steady-state value
76Inflation targeting at low, positive, inflation
Composite disturbances
77(No Transcript)
78Evolution of money supply
The only exogenous variables in the system are
the natural interest rate
nominal rate consistent with inflation target
79MICROFOUNDED CAGAN-SARGENT PRICE LEVEL
DETERMINATION UNDER MONETARY TARGETING
FLEX-PRICE, COMPLETE-MARKETS MODEL
80Complete Markets
Value of portfolio with payoff D
price kernel
81Interest coefficient for riskless asset
Riskless Portfolio
82Budget Constraint
Where T is the transfer payments based on
the seignorage profits of the central bank,
distributed in a lump sum to the representative
consumer
83No Ponzi Games
For all states in t1
For all t, to prevent infinite c
The equivalent terminal condition
84Lagrangian
85Transversality condition
Flow budget constraint
86Market Equilibrium
Market solution for the transfers T
87Monetary Targeting BC chooses a path for M
Fiscal policy assumed to be
Equilibrium is
S.t. Euler-intertemporal condition condition FOC-i
tratemporal condition TVC Constraint
For given
88We study equilibrium around a zero-shock steady
state
89Derive the LM Curve
From the FOC
At the steady state
90Separable utility
Define
The hat variables are proportional deviations
from the steady state variables.
91Similar to Cagans semi-elasticity of money
demand
92We log-linearize around
zero inflation
define
Log-linearize the Euler Equation and transform
it to a Fisher equation
Elasticity of intertemporal substitution
g is the twist in MRS between m and c
93Add the identity
We look for solution given exogenous shocks
94Solution of the system
This is a linear first-order stochastic
difference equation ,where,
Exogenous disturbance (composite of all shocks)
95given
There exists a forward solution
From which we can get a unique equilibrium value
for the price level
This is similar to the Cagan-Sargent-wallace
formula for the price level, but with the
exception that the Lucas Critique is taken care
of and it allows welfare analysis.
96I. Interest Rate Targeting based on exogenous
shocks
Choose the path for i specify fiscal policy
which targets D
Total end of period public sector liabilities.
Monetary policy affects the breakdown of D
between M and B
No multi-period bonds
Beginning of period value of outsranding bonds
End of period, one-period risk-less bonds
97Steady state (around
)
fix
98PRICE LEVEL IS INDETERMINATE
Real balances are unique
Future expected inflation is unique
Is unique
But, neither
Can uniquely be determined!
99To see the indeterminancy, let denote
solution value
v is a shock, uncorrelated with (sunspot), the
new triple is also a solution, thus
Price level is indeterminate under the interest
rule!
100II. Wicksellian Rules interest rate is a
function of endogenous variables (feedback rule)
Vcontrol error of CB
Fiscal Policy
Exogenous
Endogenous
101Steady State
Log-linearize
102We can find two processes
Add the identity
1031), 2) and 3) yield
P is not correlated to the path of M money
demand shocks affect M, but do not affect P the
LM is not used in the derivation of the solution
to P.
104FEATURES
- Forward looking
- Price is not a function of i rather , a function
of the feedback rule and the target - suppose
105Additionally
Price level instability can be reduced by raising
, an automatic response.
106Note, also that
, reduces the need for accurate observation of
, almost complete peg of interest rate
107The path of the money supply
By using LM, we can still express
But we must examine existence of a
well-defined demand for money. Theres possibly
liquidity trap
108III. TAYLOR (feedback) RULE
Assume
109Taylor principle
Is predetermined
110Transitory fluctuations in
Create transitory fluctuations in
Permanent shifts in the price level P.
111Optimizing models with nominal rigidities
112(No Transcript)
113(No Transcript)
114First Order Conditions
115Firms Optimization
Nominal
Real
116Natural Level of Output
117Log-linearization of real mc
Partial-equilibrium relationship?
118where
Elasticity of marginal product of labor wrt output
Elasticity of wage demands, wrt to output
holding marginal utility of income constant
119ONE-PERIOD NOMINAL RIDIGITY
Same as before, except for
Y need not be equal to the natural y
120A Neo-Wicksellian Framework
THE IS
Ct consumption aggregate
gross rate of increase in the Dixit-Stiglitz
price index Pt
121Equilibrium condition
A log-linear approximation around a deterministic
steady state yields the IS schedule
gcrowding out term due to fiscal shock
122Effect on fiscal shock on C
Equivalent to the fiscal shock
123New Keynesian Phillips Curve
Deviation of natural output due to supply shock
Demand determined output deviations
Taylor Rule
Inflation target
124Output gap
3-EQUATION EQUILIBRIUM SYSTEM
Proportion of firm that prefix prices
IS-curve involves an exogenous disturbance term
125INTEREST RULE AND PRICE STABILITY
THE NATURAL RATE OF INTEREST
126Percentage deviation of the natural rate of
interest from its steady-state value
127Inflation targeting at low, positive, inflation
Composite disturbances
128(No Transcript)
129Evolution of money supply
The only exogenous variables in the system are
the natural interest rate
nominal rate consistent with inflation target