American Economic Association

Asset Prices under Habit Formation and Catching up with the Joneses
Author(s): Andrew B. Abel
Source: The American Economic Review, Vol. 80, No. 2, Papers and Proceedings of the
Hundred and Second Annual Meeting of the American Economic Association (May, 1990), pp.
38-42
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/2006539 .
Accessed: 03/02/2011 06:19

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=aea. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review.

http://www.jstor.org
 AGGREGATEASSET PRICINGt

Asset Prices under Habit Formation and Catching up

with the Joneses

By ANDREWB. ABEL*

This paper introducesa utility function tion in period t-1 and Ct-1 is aggregate
that nests three classes of utility functions: consumptionper capita in period t -1. If
1) time-separable utility functions; 2) y = 0, then v,-- and the utility functionin
"catchingup with the Joneses"utility func- (1) is time separable.If y > 0 and D = 0, the
tions that dependon the consumer'slevel of parameter v, depends only on the lagged
consumptionrelativeto the laggedcross-sec- level of aggregateconsumptionper capita.
tional averagelevel of consumption;and 3) This formation is the relativeconsumption
utility functions that display habit forma- model or "catchingup with the Joneses."1
tion. Incorporatingthis utility functioninto Finally, if y > 0 and D = 1, the parametervt
a Lucas (1978) asset pricing model allows depends only on the consumer'sown past
calculationof closed-formsolutions for the consumption.This formulationis the habit
prices of stocks, bills and consols underthe formationmodel.
assumptionthat consumptiongrowthis i.i.d. Considerthe effectson utilityof a change
Then equilibriumasset prices are used to in an individual's consumptionat date t,
examinethe equitypremiumpuzzle. holding aggregateconsumptionunchanged.
Substituting(2) into (1) and then differenti-
I. TheUtilityFunction ating with respectto c, yields

At time t, each consumerchoosesthe level (3) dU,/dc, = ujc(c, v,)

of consumption,C1, to maximizeE,(U,) where
Et{ } is the conditionalexpectationoperator + /u,( ct+l, Vt+) yDv,+1/ct.
at time t and the utilityfunctionis givenby
00 Suppose that the period utility function
(1) U,- u(c,, v,) has the followingisoelasticform
juct+i,v,+i)
j=0
(4) u(c,,v,) = [ct/v]l a/(l-a), a>0.
where vt+j is a preferenceparameter.Sup-
pose that the preferenceparameterv, is spec- When y = 0, the utility functionin (4) is
ified as the standardconstant relativerisk-aversion
utility function and a is the coefficientof
(2) , -
c,D1C1J?D]Y y?O0 and D2O relativerisk aversion.Moregenerally,utility
dependson the level of consumptionrelative
where ct_1 is the consumer'sown consump-
'The phrase "catching up with the Joneses," rather
than "keeping up with the Joneses," reflects the as-
tDiscussants: Phillippe Weil, Harvard University;
sumption that consumers care about the lagged value of
Narayana Kocherlakota, Northwestern University;
aggregate consumption. The April 1989 version (p. 10)
Stanley Zin, Carnegie Mellon University.
of Jordi Gali (1989), but not the September revision,
*Department of Finance, Wharton School of the examines the utility function u(ct, Ct) = [1/(2 - I -
University of Pennsylvania, Philadelphia, PA 19104. I y)]cl - (ct/C,)- (- -) and shows that when ,8 = 1, asset
thank Mike Perigo for helpful discussion and excellent pricing will be equivalent to an economy without con-
research assistance. sumption externalities and with log utility.
38
VOL. 80 NO. 2 A GGREGATE ASSET PRICING 39

to some endogenous time-varying bench- counted utility. Suppose that a consumer

mark vP.2 Under the isoelastic utility func- reduces c, by 1 unit, purchases an asset with
tion in (4), the expression for dU8/dc, in (3) a gross rate of return Rt+1, sells the asset in
becomes period t +1, and increases ct+1 by Rt+1
units. The equilibrium rate of return Rt+1
(5) d8L/dct must satisfy

=[1-ftyD( Ct+llct ) -O( Pt/ t+ 1) ]" (7) Et { -( dUtldc,)

X ( ctl Pt) (ilct).

+ Rt+l( dUt+l/dct+1)} =0.

II. Equilibrium Equation (7) can be rewritten as

Let y, be the amount of the perishable (8) Et({fRt+( dUt+l/dct+1)

consumption good per capita produced by
the capital stock. In equilibrium, all output /E{t dU/act}} =1.
is consumed in the period in which it is
produced, as in Lucas. Because all con- Equation (8) is the familiar result that the
sumers are identical, ct = Ct=y in every conditional expectation of the product of the
period. Now let x y41 Yt+l/Yt be the gross intertemporal marginal rate of substitution
growth rate of output. Because ct = Ct= Yt, and the gross rate of return equals one.4
it follows that ct+1/Ct = Ct+11Ct = xt+l We can obtain an expression for (dUt+l/
Therefore, equation (2) implies that it+lvt act+ )/Et { dUt/d ct} using equation (6) to
= xY which allows us to rewrite (5) as divide d Ut+1/dct+ 1 by Et { d Ut/dct } to ob-
tain
(6) dUtldct = Ht+jPt` -lt
(9) (dbUt+l/dct+)/Etf{ dUc/at}
where Ht+1=-1-#/yDxt+xt-y(l-a)
Note that Ht+1-1 if yD = 0, which is the =[Ht+ 2/Et { Ht+ 1 }]Xty'a )t1
case for both time-separable and relative
consumption preferences.3 IV. The Priceof RiskyCapital

Ill. AssetPricing Let pts be the exdividend price of a share

of stock in period t, which is a claim to a
To calculate asset prices, let us examine a unit of risky capital. The rate of return on
consumer who considers purchasing an asset stock is R s 1 pts+1+ yt+l)pts- Let wt
in period t and then selling it in period t + 1. p_/yt be the price-dividend ratio. Therefore,
If asset prices are in equilibrium, this pair of Pt = wtYt and Pts+i= Wt+iyt+i so that
transactions does not affect expected dis-
(10) Rst+l =( + Wt+i)Xt + IWt-

2George Constantinides (1988), Jerome Detemple

Substituting (10) into (8) yields
(1989), John Heaton (1989), and Suresh Sundaresan
(1989) also examine asset prices in the presence of habit (1 1) wt= aEt{ (1 + wt+l) xt+ L
formation. James Nason (1988) includes a time-varying
benchmark level of consumption that differs from habit x ( dUt+lldct+,)IEt { dut/dct }
formation in that it is independent of an individual
consumer's own consumption.
3A sufficient condition for dU1/d c > 0 when y = D
= 1 (habit formation ) is 1 + ln f3/ln(max{ x }/min{ x)
< a < 1 + lnf3/ln(min{ x }/max{ x }). For 3= 0.99 and 4In the conventional time-separable formulation of
the 2-point distribution in Table 1, the sufficient condi- this problem, d Ut/I ct is known as of time t, and hence
tion is 0.858 < a < 1.142. Et { d Ut/d ct } on the left-hand side of (8) equals d Ut/dct.
40 AEA PAPERS AND PROCEEDINGS MA Y 1990

V. Bills andConsols and the price of a consol is

A one-period riskless bill can be pur- (16) pic = Qx4@/Jt,

chased in period t at a price of s,; in period
t + 1, the bill is worth 1 unit of consumption. where Q=A3q/[1-flEx{xa}].
The gross rate of return on the bill is RI1=
l/st. Substituting l/st for the rate of return Given a distribution for x, the moments of
in (8) yields x can be calculated and the three asset prices
are easily calculated. For time-separable
(12) st=f,Et{(dUt+l/dct+1) preferences (y = 0) and relative consumption
(y > 0; D = 0), we can obtain closed-form
/Et{ dUt/dc,} }. solutions (in terms of preference parame-
ters and the moments of x) for the uncondi-
A consol bond, that pays one unit of tional expected returns E{ RS}, E{ RB} and
consumption in each period, can be pur- E{Rc):
chased at an excoupon price ptc in period t.
In period t +1, the consol pays a coupon (17) E{RS} =E{x-9}
worth one unit of consumption and then
sells at a price of ptc+1i The one-period x [E{ x} +A E{ xl+O}]/A
rate of return on the consol is Rc+1
(1 + j4i+-)/ptc. Substituting RC 1 into (8)
yields (18) E{ RB} = E{ x-@}/fq

(13) pc=fEt{(1+ptc+l)(dbU+j/dct+1) (19) E{ RC} = E{ x-@} [1+ QE{ x?}]/Q.

/E,{ dU,/dc,} }. Under habit formation, unconditional ex-

pected returns can be calculated numerically
VI. I.I.D. ConsumptionGrowth using the asset prices in (14)-(16).

Suppose that consumption growth x,+1 is VII. The EquityPremium

i.i.d. over time. In this case, we can obtain
explicit solutions for the prices of stock, Rajnish Mehra and Edward Prescott
bills, and consols. The price-dividend ratio (1985) report that from 1889 to 1978 in the
wt is United States, the average annual real rate
of return on short-term bills was 0.80 per-
(14) wt= A x,l, cent and the average annual real rate of
return on stocks was 6.98 percent. Thus
where =-y(a-1) the average equity premium was 618 basis
points. They calibrated an asset pricing
A -PE{ xl-} [1 - /3yDEt X(l-a)(l-Y) }J model with time-separable isoelastic utility
to see whether the model could deliver un-
/[1-E{ E (l-a)(l-Y) }] conditional rates of return close to the his-
torical average rates of return on stocks and
Jt Et{ Ht+1} -1-,ByDE{ xl-a} x, bills. They used a 2-point Markov process
for consumption growth with E{ x, } = 1.018,
The price of a one-period riskless bill is Var{x,} = (0.036)2, and correlation (x,,
xt1) = -0.14. For values of the preference
(15) s,= q/xt'/Jt, parameters that Mehra and Prescott deemed
reasonable, the model could not produce
where more than a 35 basis point equity premium
(E{Rs}-E{RB}) when the expected risk-
q Et x} a - /yDE{ x1-a} E{ xo-a} less rate, E{ RB), was less than or equal to 4
VOL. 80 NO. 2 A GGREGATE ASSET PRICING 41

TABLE 1-UNCONDITIONAL EXPECTEDRETURNS premium is 463 basis points and the un-
13= 0.99; E{x } = 1.018; VAR{ X} = (0.036)2 conditional riskless rate is 2.07 percent per
ax Stocks Bills Consols year. Although the unconditional expected
returns on stocks and bills are much closer
A. Time-separable preferences (y = 0) to their historical averages, the conditional
0.5 1.93 1.87 1.87
[1.93] [1.87] [1.87]
expected rates of return (not reported in the
1.0 2.83 2.70 2.70 table) vary too much. For the 2-point distri-
[2.83] [2.70] [2.70] bution for x, the standard deviation of
6.0 10.34 9.52 9.52 E,{R'+1} is 17.87 percent when a=6. This
[10.33] [9.51] [9.51] unrealistic implication of the model poses a
10.0 14.22 12.85 12.85
[14.13] [12.72] [12.72] challenge for future research.
B. Relative consumption (y = 1; D = 0) Panels A and B report unconditional rates
0.5 2.80 2.76 2.73 of return for a lognormal distribution with
[2.80] [2.76] [2.73] E{x} =1.018 and Var{x} = (0.036)2. For
1.0 2.83 2.70 2.70
[2.83] [2.70] [2.70] the parameter values reported, it makes no
6.0 6.70 2.07 5.84 substantial difference for expected returns
[6.72] [2.06] [5.86] whether the growth rate is lognormal or has
10.0 14.73 1.59 13.16 a 2-point distribution.
[14.95] [1.55] [13.32]
Panel C presents the unconditional ex-
C. Habit formation (y = 1; D = 1)
0.86 33.56 4.53 35.25 pected rates of return under habit formation.
0.94 6.83 3.48 7.44 The expected rates of return on both long-
1.00 2.83 2.70 2.70 lived assets (stocks and consols) are ex-
1.06 8.43 1.93 7.40 tremely sensitive to the value of a. Under
1.14 38.28 0.93 35.16
logarithmic utility (a = 1), the expected rates
of return are the same as under time-sep-
arable preferences and relative consumption.
percent per year. This result is the equity However, with a = 1.14, the expected rates of
premiumpuzzle. return on stocks and consols are both greater
Table 1 reports the unconditional ex- than 35 percent.
pected rates of returnon stocks, bills, and Further research using the utility function
consols under the assumption that xt is introduced in this paper will explore the
i.i.d., E x) = 1.018 and Var{x) = (0.036)2. implications of other settings for the param-
For time-separableand relativeconsumption eters y and D. For instance, if D is between
preferences,two unconditionalexpectedre- zero and one, the utility function would con-
turns are reportedin each cell: the first is tain elements of both catching up with the
calculatedundera 2-pointi.i.d. distribution; Joneses as well as habit formation. Also the
the second, shown in brackets,is calculated assumption of i.i.d. consumption growth
undera lognormaldistributionfor x. rates can be relaxed, and asset prices can
Panel A of Table 1, which reports the then be analyzed numerically.
unconditionalexpectedratesof returnunder
time-separablepreferences,displays the eq-
uity premiumpuzzle. Although E{ Rs'} in- REFERENCES
creases as a increases from 0.5 to 10.0,
E{ RB) also increases.The equitypremium, Constantinides,GeorgeM., "Habit Formation:
E{Rs}-E{RB}, does not come anywhere A Resolution of the Equity Premium Puz-
close to the 600-pointhistoricalaverage.In- zle," University of Chicago, October 1988.
cidentally, the unconditionalexpectedrates Detemple,Jerome,"General Equilibrium with
of return of bills and consols are exactly Time Complementarity: Consumption, In-
equal undertime-separablepreferences. terest Rates, and the Equity Premium,"
Panel B reports the unconditional ex- Graduate School of Business, Columbia
pected rates of returnin the relative con- University, September 1989.
sumption model. For a = 6, the equity Gali, Jordi, "Keeping up with the Joneses
42 AEA PAPERS AND PROCEEDINGS MA Y 1990

Consumption Externalities, Portfolio Equity Premium:A Puzzle,"Journal of

Choice and Asset Prices," mimeo., Gradu- Monetary Economics, March 1985, 15,
ate School of Business, Columbia Univer- 145-61.
sity, September 1989. Nason, James, "The Equity Premium and
Heaton,John, "An Empirical Investigation of Time-Varying Risk Behavior," Finance
Asset Pricing with Temporally Dependent and Economics Discussion Paper Series,
Preference Specifications," MIT, Decem- No. 11, FederalReserveBoard,February
ber 1989. 1988.
Lucas, RobertE., Jr., "Asset Prices in an Ex- Sundaresan,Suresh M., "Intertemporally De-
change Economy," Econometrica,Novem- pendent Preferencesand the Volatilityof
ber 1978, 46, 1429-45. Consumptionand Wealth,"Reviewof Fi-
Mehra, Rajnishand Prescott,EdwardC., "The nancial Studies, 1989, 2, 73-89.