1
Consumption
Lifetime utility:
T 1
U u (Ct ) with u() 0, u() 0
t 1 (1 )
t 1
Budget constraint:
T 1 T 1
t 1
C t A0 Y
t 1 t
t 1 (1 r ) t 1 (1 r )
Lagrangian:
T
1 T
1 T
1
L t 1
u (Ct )
A0 Y
t 1 t
C
t 1 t
t 1 (1 ) t 1 (1 r ) t 1 (1 r )
(1 ) t 1
F.O.C.: u (C t )
(1 r ) t 1
If r=, then u(Ct) is constant over time, so C is
constant over time
If >r, then u(Ct) is rising over time, so C is falling
over time
If r>, then u(Ct) is falling over time, so C is rising
over time
2
2 special cases:
1) u(C) = ln[C] and T=
1 (1 ) t 1
Ct (1 r ) t 1
1 (1 r ) t 1
Ct
(1 ) t 1
1 1 1
t 1 (1 )
t 1
A0
t 1 (1 r )
t 1
Yt
1
1
A0 Yt
1 t 1 (1 r )
t 1
(1 r ) t 1
1
Ct
(1 ) t 0
A
t 1 (1 r )
Y
t 1 t
2) r==0, and T is finite
1 T
Ct A0 Yt
T t 1
Permanent Income Hypothesis (Milton Friedman ): consumption
depends on your permanent income and rises one-for-one with
permanent income.
C=YP
Life-cycle model (Franco Modigliani): people plan consumption
over their lifetimes to maximize the present value of their utility
save in periods in which income is relatively high and dissave in
periods in which income is relatively low
3
Transitory income: YT=Y-YP
Savings: S = Y C
Interpreting a simple consumption function
If we looked at a consumption function over long time
periods, consumption is approximately proportional to
income
However, if we looked at consumption at shorter time
horizons or in a cross section of individuals, consumption can
be expressed by the relationship C=a+bY, where a>0 and
0<b<1.
Suppose we estimate Ci=a+bYi+ei.
Cov (Y , C ) Cov (Y P Y T , Y P ) Var (Y P )
b
Var (Y ) Var (Y P Y T ) Var (Y P ) Var (Y T )
a C bY Y P
b(Y P
Y T ) (1 b)Y P
4
The random walk hypothesis for consumption (Robert Hall)
If future income is uncertain, then consumption should follow a
random walk (perhaps with drift)
random walk: xt=xt-1+et
random walk with drift: xt=xt-1++et
T
1
E[U ] Et u (Ct )
t 1 (1 )
t 1
T
1 T
1
s.t. t 1
Ct A0 Y
t 1 t
t 1 (1 r ) t 1 (1 r )
Suppose consumption in period 1 falls by the amount dCt. This
reduces utility in period 1 by the amount u(Ct)dCt.
This means that consumption in period 2 can be raised by (1+r) dCt,
and the utility of this (in terms of todays dollars) is
1 r
E t [u (C t 1 )]dC t
1
If the individual had previously been maximizing his or her
utility,
1
E t [u (C t 1 )] u (C t ) Euler equation
1 r
Suppose the discount rate = the interest rate
Et [u (Ct 1 )] u (Ct )
Suppose also that utility is quadratic with the following form:
5
a 2
u (C t ) C t Ct
2
6
What determines changes in consumption over time?
1
C1 A0 E1 [Yt ]
T
T t 1
1
A1 E 2 [Yt ]
T
C2
T 1 t 2
1 E1 [Yt ]
T T T
C2 A
0 1 1
Y C E [Y ] E [Y ]
T 1 t 2
1 t t 2 2 t
t 2
1
TC1 C1 E2 [Yt ] E1[Yt ]
T T
C2
T 1 t 2 t 2
T
since TC1 A0 Y1 E1[Yt ]
t 2
1 T
E 2 [Yt ] E1[Yt ]
T
C 2 C1
T 1 t 2 t 2
7
Tests of the Permanent Income/Random Walk
Hypothesis:
Hall (1978)
Flavin (1981)
regressed current income on lagged income:
yt 1 1 yt 1 2 yt 2 ... p yt p 1t
also ran the following regression for change in
consumption:
ct 2 k ( y t 1 1 y t 1 2 y t 2 ... p y t p )
0 y t 1 y t 1 2 y t 2 ... p 1 y t ( p 1) 2t
term in ( ) is innovation in income. I.e., the
difference between current income and amount
predicted from past income
represents how much an innovation in current
income raises expectations of lifetime income
k represents how much a rise in lifetime income
should raise current consumption
represents normal change in consumption
8
Found that hypothesis that 0=1=p-1=0 could
be rejected at 0.5% level.
Estimated coefficients: 0=.355, 1=.071,2=.049
9
Campbell and Mankiw (1989)
hypothesized that a fraction, , of consumers spend
their current income and the remaining (1- of
consumers behave according to Halls theory
Ct Ct 1 (Yt Yt 1 ) (1 )et
Ct Yt (1 )et
C t Z t t
where Z t Yt
t (1 )et
et represents the change in consumers estimates of
their permanent income between t-1 and t. Treated as
a random error.
Test for Halls model is test that =0
Estimated equation: C t Z t ~t
C t Z
t (Z t Z
t ) t
C t Z
t ~
t
~
where ( Z Z )
t t t t
Shea (1995) examined change in spending by workers covered
by long-term union contracts
Souleles (1999) examined income tax refunds
10
Hsieh (2003) examined payments to Alaska residents from oil
royalties
Parker, Souleles, Johnson, and McClelland (2013) examined
consumer spending resulting from stimulus package of 2008.
Single taxpayers generally received $300-$600, and couples
generally received $600-$1200. In addition, households received
$300 per child who qualified for the child tax credit. Timing of
receipt was based on last two digits of SSN.
11
Liquidity Constraints
Test for liquidity constraints:
1) Zeldes (1989)
1 rit
Euler equation: u (C it )
1 i
E t [u (Ci ,t 1 )]
If an individual is liquidity constrained in period t,
then
1 rit
u (Cit ) Et [u (Ci ,t 1 )] it
1 i
2 period model
1
Max u (Cit ) Et [u (Ci , t 1 )]
1 i
s.t. Ci , t 1 ( Ait Yit Cit )(1 rit ) Yi , t 1
s.t. Cit Ait Yit
1
L u (Cit ) Et [u (( Ait Yit Cit )(1 rit ) Yi , t 1 )]
1 i
it ( Ait Yit Cit )
dL 1
0 u (Cit ) Et [u (( Ait Yit Cit )(1 rit ) Yi , t 1 )(1 rit )]
dCit 1 i
it
1 rit
u (Cit ) Et [u (Ci , t 1 )] it
1 i
12
2) Shea (1995)
Time-inconsistent preferences (hyperbolic discounting)
Consumption and Risky Assets
Assume individual can reduce consumption in period t by
an infinitesimal amount and use it to buy a risky asset, i.
Let P represent the price of the asset and D , D ,
t
i i
t 1
i
t 2
represent the future stream of payoffs, which are uncertain.
13
1
u (Ct ) Pt i Et u (Ct k ) Dtik for all i
k 1 (1 )
k
Suppose assets are held for one period.
Di
Then rti1 t i 1 1
Pt
Et (1 rt i1 )u (Ct 1 )
1
u (Ct )
1
for all i
u (Ct )
1
Et 1 rti1 Et u(Ct 1 ) Cov 1 rti1 , u(Ct 1 )
1
for all i
With quadratic utility, u(Ct 1 ) 1 aCt 1
u (Ct )
1
Et 1 rti1 Et u(Ct 1 ) aCov1 rti1 , Ct 1
1
Note that the risk of the asset does not appear in the above
equation
Optimal holding of an asset depends on the correlation of
the assets return and consumption
Home country bias
Consumption CAPM
14
Et 1 rti1 (1 )u(Ct ) aCov1 rti1 , Ct 1
1
Et u(Ct 1 )
The higher an assets covariance with consumption, the
higher its expected return must be.
Let rt 1 represent the rate of return on a risk-free asset
(1 )u (Ct )
1 rt 1
Et u (Ct 1 )
aCov1 rti1 , Ct 1
Et rti1 rt 1
Et u (Ct 1 )
Equity Premium Puzzle
Let represent the coefficient of relative risk aversion,
assuming CRRA utility function,
Ct1
U (Ct ) .
1
Then it can be demonstrated that
E[r i ] E[r ] Cov(r i r , g c ) ,
where gc represents the growth in consumption.
15
From 1890-1979, the difference between the average return
on the stock market and the return on short-term
government securities averaged about 6%, and the
covariance between consumption growth and the excess
return on the market was 0.0024.
Implies that =25.