Wiley The Scandinavian Journal of Economics

Inflation and Unemployment in General Equilibrium Author(s): Guillaume Rocheteau, Peter Rupert and Randall Wright Source: The
Scandinavian Journal of Economics, Vol. 109, No. 4, Macroeconomic Fluctuations and the Labor Market (Dec., 2007), pp. 837-855 Published by: Wiley on behalf of The Scandinavian Journal of Economics Stable URL: http://www.jstor.org/stable/25195319 . Accessed: 01/10/2013 03:27
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
.
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 support@jstor.org.
Wiley and The Scandinavian Journal of Economics are collaborating with JSTOR to digitize, preserve and extend access to The Scandinavian Journal of Economics.
http://www.jstor.org
This content downloaded from 203.170.79.82 on Tue, 1 Oct 2013 03:27:47 AM All use subject to JSTOR Terms and Conditions
Scand. J. of Economics 109(4), 837-855, DOI: 10.HH/j.1467-9442.2007.00511.x
2007
Inflation and Unemployment

General
Guillaume
Federal Reserve
in
Equilibrium*
Rocheteau
Bank, Cleveland, OH 44114, USA
guillaume.rocheteau@clev.frb.org
Peter Rupert
of California, University rupert@econ.ucsb.edu Santa Barbara, CA 93106, USA
Randall Wright
University of Pennsylvania, Philadelphia, PA 19104, USA rwright@econ.upenn.edu
Abstract
When ployed, labor as some agents randomly unem outcomes with there exist efficient is indivisible, in Rogerson this idea into the modern (1988). We integrate theory of monetary some trade occurs where in centralized markets and some in decentralized markets, and Wright with a general model of unemployment (2005). This delivers equilibrium foundations. We show that the implied relation explicit microeconomic can be positive or negative, on simple pref and unemployment depending curve provides a long-run, Our Phillips trade-off for monetary exploitable, that the optimal inflation; ?12; policy is the Friedman rule.
exchange, as in Lagos and money, between erence policy; Keywords'. JEL
inflation conditions.
it turns out, however, curve; ?40;
Phillips
unemployment ?13
classification:
?52;
I. Introduction The has been well known at least since the work of Rogerson following efficient in economies with indivisible allocations labor generally (1988): have some agents, chosen at random, unemployed, while others are em can be sup ployed, even if they are ex ante identical; and these allocations as we discuss where As trade lotteries. agents ported competitive equilibria
to this project, we thank Karl Shell, Bob Hall, Ken *For input on previous work related Christian at the Todd Keister, and seminar participants Burdett, Aruoba, Boragan Helwig, Cleveland Fed, Bank of Canada, Fed, Philadelphia Penn, Penn State, Princeton, LSE, UCLA, of Singapore, Simon Fraser, Essex, and the Cana Cornell, Notre Dame, National University at UBC. We thank the NSF Theory meetings here are those of the authors and not necessarily expressed or the Federal Reserve Bank of Cleveland System. dian Economic for research those of support. The views the Federal Reserve
? The editors of the Scandinavian Journal of Economics 2008. Published by Blackwell Publishing, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street,Maiden, MA 02148, USA.
838
G Rocheteau,
P. Rupert and R. Wright
in detail below, another interesting feature of these economies emphasized in Rocheteau, (2006) is that agents act as if they Rupert, Shell and Wright have quasi-linear utility. It turns out that one can use this result to con a struct fairly general yet very tractable model of monetary exchange, using search theory, where (as in Lagos and Wright, 2005) some trades occur in centralized markets and some in decentralized markets. To understand tractable is the this, note that what makes Lagos-Wright of quasi-linear utility, which implies agents exiting the central all hold the same amount of money, regardless of their histories to embrace quasi if one is willing interior solutions). Thus, (assuming one can in the avoid having to track the distribution of money linear utility, in Rocheteau market as a state variable.1 The observations decentralized as long as we have et al. (2006) allow one to dispense with quasi-linearity: assumption ized market indivisible for that matter, any indivisible commodity?identical labor?or, results concerning the money distribution hold for any utility function (again to using there are some advantages interior solutions). However, assuming indivisible labor instead of quasi-linear utility as a building block for mon etary theory, including the fact that it generates unemployment. In this paper we take seriously the implications of indivisible-labor mod In els with money for the relationship between inflation and unemployment. other words, we study the Phillips curve in general equilibrium. This seems to be a natural exercise. In addition to Rogerson (1988), many well-known papers adopt the indivisible-labor model, (1985), Cooley including Hansen and Eichenbaum and Hansen (1992), Kydland (1994), (1989), Christiano and Wallenius Prescott, Rogerson (2006). (2006) and Sargent and Lungqvist or with issues either without money These papers analyze macroeconomic some in ad hoc via added cash-in-advance way (e.g. money constraints). and inflation in a We want to study the relation between unemployment of with relatively explicit descriptions model with microfoundations?i.e., the frictions that make money essential.2 The main goal is to derive precise analytic results showing how the rela inflation and unemployment tion between depends on primitives. We prove on that the Phillips curve can have a positive or negative slope depending
1 one has For monetary models that are much less tractable because see Green and Corbae relevant distribution, and Zhou (1998), Camera Zhu were (2003, also 2005) simple, {0, and Molico (2006). Earlier search-based models, to keep track of the (1999), Zhou (1999), such as Kiyotaki and could only to achieve
Wright (1989, 1993),Aiyagari andWallace (1991), Shi (1995) or Trejos andWright (1995),
hold m 2 An 1} units but only because An of money. the issue by assuming agents they avoided that uses alternative large families approach
tractability is provided by Shi (1997). See also Faig (2005).

alternative unemployment sarides (1994), are advantages ? Menzio and Wright taken up in Berentsen, approach, into monetary the search-based model theory using As we discuss labor model. instead of the indivisible to each approach. is to integrate (2007), of Mortensen and Pis in the conclusion, there
The editors of the Scandinavian Journal of Economics 2008.
Inflation
and unemployment
in general
equilibrium
839
the utility function in a simple and natural way. The intuition is straight forward. To consider one version of our results, suppose the economy has two sectors, one of which uses cash relatively Inflation is ob intensively. a tax on activity sector. If goods traded in in the cash-intensive viously that sector are substitutes for those traded in the other sector, inflation re of the former and increases consumption duces consumption of the latter. If those latter goods are relatively labor intensive, employment must increase in equilibrium. Hence, when the goods in the two sectors are substitutes, inflation reduces unemployment. when the goods are com Symmetrically, inflation increases plements unemployment. Our inflation-unemployment is based on rudimentary trade-off public finance considerations, and does not depend on any complicated features like sticky wages or prices, irrational expectations, imperfect information, etc. Thus we conclude that one does not have to work very hard to generate an interesting relation between inflation and unemployment. in this Now, is paper, we do not take a stand on what the relation is in the data?that an entirely different project. Here we focus on providing a simple charac terization in theory. Also, note that our trade-off is a long-run trade-off in the sense that it does not depend on features like stickiness or imperfect that are likely to be important only in the short run. And it is it fully exploitable by monetary policy: under conditions we make precise, as to permanently isfeasible reduce unemployment by increasing inflation, (used to?) think. We prove, however, that it is optimal to reduce Keynesians as Friedman always said.3 inflation to a minimum, information
II. Basic Assumptions

Time is discrete. There is a [0,1] continuum of agents who live forever. There are two types of markets in which these agents interact. One is a or CM. The other is a decentralized mar frictionless centralized market, that together make money essential: ket, or DM, with two main frictions
that our results are reminiscent of Stockman shows that suggested (1981), who in the long run when on in reduces there is a cash-in-advance constraint capital on consumption to a cash-in-advance we vestment constraint Indeed, (as opposed only). to those that one could derive results very in some of our proposi similar acknowledge referee inflation tions in indivisible-labor goods allows models are some assuming as long as one subject in a reduced-form is introduced fashion money by to a cash-in-advance are not, constraint and some goods so that one can compare the cases where these preferences where
3 A
general are complements and when and allows different labor intensities goods they are substitutes, see no reason to take such a short cut, however, We in production. since it is no harder and it generates one derives the role of money additional from first principles. insights when are wed we emphasize But for those who to reduced-form that our main economic models, results also hold in that context. ?
840
G. Rocheteau,
as detailed below, and anonymity, which problem, credit precludes private arrangements.4 We are interested in equilibria where is valued?i.e., the price of M is positive in the CM at every date? money and choose dollars to be the unit of account. The stock of money evolves = z to M ? where indicates the value of variable any (\ +j)M9 according next period. New money is injected (or withdrawn if 7 < 0) via lump-sum transfers (or taxes) in the CM, to neutralize the fiscal effects of monetary policy.5 It is easy to allow a general vector of consumption goods x R+ and e e in the CM; see our working paper, Rocheteau, endowment Rupert R+ and Wright In order to reduce notation, however, we assume a (2007). = 0. This single CM good x and e good is produced by firms using labor as h. For an individual, labor is indivisible: he {0, 1}. Hence, or in trade randomized bundles, lotteries, agents consumption the DM there is a different good, q, that is not produced, but an endowment one CM in an interval encompassing q. Utility a at start where shock the is realized j preference vJ(q,x,h), is standard, the CM. In agents have and DM is of the DM, after (x,h) but before q is chosen. This timing, and in particular the before that the shock is revealed after but (x,h) preference assumption that q is chosen, is not especially important, but we like the interpretation the CM closes before the DM opens. The related monetary literature assumes utility is separable between (jc,h) and q, but for reasons that will become clear we do not want to restrict attention to this case. Indeed, papers following Lagos and Wright (2005) assume utility is linear in either x or h, but as will become clear we do not it would be straightforward need to restrict attention to this case. Although or endowment to allow general preference shocks, it suffices here to as sume the following: with probability a the utility function is vH, and with for all I?a it is vL, where dvH(q,x,h)ldq> probability dvL(q,x,h)/dq, (q,x,h), which generates gains from trade similar to, for example, Berentsen is that there is no and Rocheteau (2003). The key point for our purposes
and Wright (1998), Wallace (2001), Corbae, Temzelides (2003) or Alipran Puzzello for formal discussions of essentiality and the role of (2007) et al. is especially since they explicitly consider models with relevant anonymity. Aliprantis and decentralized both centralized meetings. 5 out to us, under As Todd Keister the alternative that increasing M pointed assumption an means in in this increase reduces model, 7 government always increasing spending, See Kocherlakota tis, Camera and on employment is the wealth effect of fiscal by emphasized unemployment?this policy While Christiano and Eichenbaum this is and potentially (1992). quantitatively interesting, currency, relevant, we want to focus on the effects of monetary policy on the return to holding so we hold government constant. Note, rather than the pure wealth effect, spending by the way, there on ? in case there is no wealth supply. is any confusion, quasi-linear but effect on money demand, labor models utility or indivisible a wealth there is still most definitely imply effect 4
a double-coincidence
labor
Inflation
and unemployment
in general
equilibrium
841
so employment in the DM?it is a pure exchange market?and production in worked the CM. is unambiguously hours given by Trade in the DM is bilateral, and so we have to say how people meet. as in one could consider a more general matching technology, Although = a et al. (2007), here we set and assume every agent that Rocheteau \ draws vH is matched with one that draws vL. We call the former a buyer and the latter a seller, since there is a deal to be done where the agent with vL transfers some q to the one with vH in exchange for some cash. the frictions in the DM imply an essential role for some medium Generally, of exchange. We assume that the good x cannot be carried into the DM, and there is no other storable object, so that this role must be played by this paper is not about the coexistence of money money (in other words, and other assets). on utility to guarantee con We impose standard curvature conditions x one of With is indivisible labor, sumption strictly positive. obviously similar for A, and interiority of (the probability do something of) an we assume to return is which issue below. We dis agents employment count between the DM and next CM at rate ? e (0, 1), but not between the CM and DM. Let W(m) denote the CM value function, which depends only on money since in all other respects agents are identical in this balances, denote the DM value function, which, market. Let V(m,x,h) in addition to m, also depends on (x,h) since these are given when one enters this market. This completes our description of the basic model. In the next four sections we analyze the CM, we analyze the DM, we put them together to define equilibrium, and we discuss the implications for the relation between inflation and unemployment. cannot
III. The CM
As we said, going back to Rogerson in models with indivisible (1988), labor agents trade lotteries. Thus, in the CM, given m, an agent chooses ? {t, x\, jco, rh\, mo), where i is the probability of employment (i.e., of h 1), if employed, and xq and mo while x\ and rh\ are consumption and money are consumption if unemployed. and money The problem is W(m)= max _
(l,x\,xo,mi,m0)
{?V(mx,
xx, 1) +(1
-l)V(m0,
jc0, 0)},
(1)
s.t. l{px\ where fer,
+mi)
(l
i)(pxo + mo) <wt
+ m + j M + A,
the wage, 7M trans the lump-sum money in dollars. As is standard, income, all measured of working i. What agents effectively get paid for selling a probability is slightly non-standard here is that they do not derive utility directly in p is the price and A dividend
? The editors of the Scandinavian Journal of Economics 2008.
of x, w
842
G. Rocheteau,
this market;
occur.
instead
they
take
(m,x,h)
to the DM,
where
other
events
A few additional comments are in order concerning (1). First, as is well an individual's choice will generally be contingent known in related models, on his employment status A, but if V is separable between x and h then and if V is separable between m and h then m\=mo.6 x0=xi, Second, a model et al. (2006) actually derive problem with from Rocheteau (1) standard Arrow-Debreu markets, and no lotteries, where agents trade state contingent commodity bundles [x(s), h(s), m(s)] and s represents a sunspot; for simplicity we start directly with lotteries.7 Finally, although (1) generally et al. (2006) does not have a quasi-concave function, Rocheteau objective conditions hold. show there is a unique solution and the second-order to reduce the notation we assume this is very easy to generalize, Although so that an aggregate production function that converts h into x one-for-one, for (1) can be the Lagrangian and A = 0. Hence, in any equilibrium w =p written W(m) = F(mi,xi,l)+(l-?)F(w0,JCo,0)
+Ar.-^1-(i-ô+,w+7M-î-(i-ôi
where
an interior solution A is a multiplier. Assuming are see conditions first-order below), (but xh:Vx(mh,xh,h) mh : Vm(mh,xh,h) = \, = \/p, A = 0,1 h = 0,l (2) (3)
for i e (0, 1) for now
I: V(mx,xu\)-V(m0,x(),0) m+
\lxx-xQ-\
+ mx~mo\
(4)
X:l-ixx-jl-t)x,
Since tions
6 Of
lM-i^-{l-i)^=0. P
(5)
does not depend on m, (2)-(4) constitute five equa V{t?h,Xh,h) and m, that can be solved for (x\,XQ,m\, mo, A), independent of
on the under it is separable depends V(-) is an endogenous object, and whether role for One of the function money explicitly, advantage deriving lying utility vJ(q,x,h). one some discipline: it in the utility instead of just sticking function, say, is that it imposes m In clear it will be assume enter and h cannot that case, very any e.g. separably. simply or wi =mo if xi =xq below 7 it is shown how to support effi That result is based on Shell and Wright (1993), where as sunspot equilibria instead of lottery equilibria. in non-convex economies cient allocations some in this over but in have contexts, lotteries, including general, advantages Sunspots see Garratt, Keister and Shell (2004). model, they are equivalent; course, ? The editors of the Scandinavian Journal of Economics 2008.
Inflation
and unemployment
in general
equilibrium
843
under weak
regularity
conditions.8
Once
we
have
(jci,jco, wi,wo,
A), (5)
yields i as the following function of m: pxp + mp-^M-m o o? \

e=t(m)=-7r--t?~??
p(l+xo-x\)
+ mo-m\
(6)
Individuals with more money supply less labor in the CM?i.e., they work can afford the same with a lower probability?which is how everyone a key part of the economics, (jci, jco,mi, mo, A). In terms of monetary status but not result is that the choice of m h may depend on employment on m, and hence all agents take the same money holdings out of the CM on h. conditional This is similar to the basic result in the Lagos-Wright model, where all agents take the same money holdings out of the CM, and hence we get a of m in the DM. Here we get at most a two-point degenerate distribution since in general the employed and unemployed do not have the distribution, same m, although as we said above they do when V is separable between m and h. In any case, a two-point distribution is not hard to handle; the of m on (the dependence important part is to eliminate history dependence that the result in the Lagos-Wright model m). We emphasize only holds with quasi-linear utility, while here it holds for any utility function. Another result that carries over from quasi-linear models is linear: is that W(m) = that W\m) A//?, by the envelope theorem, where we already established ? is independent of m. One might say agents in the indivisible-labor model act as if they have quasi-linear utility. Summarizing: in the CM the choice of riih
? e (0, I) for Lemma 1. Assuming and W\m) = X/p are independent
all agents of m.
IV. The DM
a meeting where the values of and ms. mb arbitrary of the good, subject to d<mb over more than he has. There trade (q,d): Lagos and Wright and Waller Aruoba, Rocheteau Consider ing solutions;
8 See Rocheteau
Rocheteau
et al. ?
buyer has mb and the seller ms dollars, for Generally, they trade d dollars for q units and q<q, since neither agent can turn are several ways to determine the terms of use Nash (2005) generalized bargaining; (2007) consider several alternative bargain and Wright (2005) consider price taking, as in
need
to rule out p{x\ implies a singularity for one
only
even Actually, see Rocheteau
special in this case, our et al. (2006).
for details, but basically the regularity condition is that we (2006) ? - w = = i, or equivalently jco 1) m0 V(m \, x\, 1) V(mo, *o> 0)> which in (2)-(4). In the indivisible-labor to occur literature, this case is known see e.g. Cooper and Wright (1987) or Rogerson utilty function; (1988). results go through, but we need a different argument. Again,
844
G. Rocheteau,
P. Rupert
and R. Wright
the Lucas-Prescott search model and price posting, labor-market (1974) as in the directed search model of Shimer (1996) and Moen (1997); and Galenianos and Kircher (2006) and Dutu, Julien and King (2007) use auc that allow for some multilateral tions (in versions of the model matches, it would clearly be inter which we could easily accommodate). Although in this paper we stick to bargaining, esting to consider various alternatives, and indeed to take-it-or-leave-it offers by the buyer. The reason is the following. There is a complication here, compared to several types of meetings there are, in principle, because many models, on their employment that can occur between buyers and sellers depending an meet status in the CM?an could seller, an employed employed buyer we an etc. if But meet could seller, adopt unemployed buyer employed take-it-or-leave-it offers by the buyer, and if we also assume that sellers' = are separable, say vL(q,x,h) then the terms + G(q), F(x,h) preferences in the meeting of trade will not depend on anything except the money to allow it will be important below the buyer mb. Although we can still to and be between q, (x,h) non-separable buyers' preferences we are therefore when results sellers' get interesting preferences separable; focus on this case to facilitate the presentation.9 a take-it-or-leave-it the buyer makes Given offer, one can easily show the buyer offers all his money d = mb, and asks for that in any equilibrium the seller indifferent between the q that makes trading and not trading: holdings of
G(q-q)
where
+ ?W(msVmb) = G(q) + ?W(ms),
function next period.10 By Lemma is the CM value 1, W(m) = to reduces and the previous W(ms +mb)?W(ms) expression mb\/p, ? ? = In general, some buyers were employed while G(q) G(q q) ?mbX/p. some were unemployed in the CM, and the former have rh\ and trade for while the latter have m0 and trade for q0 = q(m0) in the DM. q\=q(rh\) But the deal is independent of the value of (xh,h) for either the buyer or the seller as well as ms. Summarizing:
the buyer has general the case where bargaining (2007), we present 2 and 3) for 6 = 1, on main could the results (Propositions power only prove case. This is somewhat here to this special the suggestion of a referee we restrict attention = 1 does it rules since 6 many however, unfortunate, extensions?e.g. interesting preclude in the out a version sellers have to pay a fixed entry cost to participate of the model where DM. 10 that but here is the idea. It is easy to check and Wright for details, See Lagos (2005) some <m* in and a buyer threshold m* is below that his iff mb all mb money spends one can show that q is increasing in mb and mb = m* Also, q=q* implies equilibrium. = with strict equality that q < q* in equilibrium, where imply below c'(q*). This will uf(q*) iff the nominal interest rate is i= 0. In Rocheteau 9, but et al. since we ? The editors of the Scandinavian Journal of Economics 2008.
Inflation and unemployment
in general
equilibrium
845
Lemma
2. The DM q(-)
the function
bargaining is the solution
solution to
is d ? mb and q = q(mb),
where
?m?\/p
G(q)-G(q-q).
(7)
Now, for an agent in the DM with (m/,, jc^, h), when has some possibly random amount of money p,
every
other agent
V(mh,xh,h) =
\{vH[q+q(mJh),xh,h]+?W(0)} + \E?{vL [q-q(p),xh,h] + ?W(m+ p)},

and with ( ), probability 1/2 he is a
since with seller.
1/2 he is a buyer probability Inserting the separable form of i/
V(mh,xh, h) = \{vH [q+q(mh\ xh, h] + ?W(0) } + lE,{F(xh,h) + G[q-q(p)] + ?W(n + p)}.

and inserting W'(-) Differentiating, arrive at the envelope conditions and q\-) fr?m Lemmas
(8)
1 and 2, we
Vm(mk,xh,h)=--\-*-?---=??
2p [
G'[q-q(mh)]
1|
(9)
j
(10)
Vx(mh,xh, h)=\v?
[q+q(mh), xh, h] + {Fx{xh, A).
V.
Equilibrium
some results. First, insert the expressions We begin by collecting for V and its derivatives in (8)?(10) plus the bargaining solution (7) from the DM into the first-order conditions (2)-{4) from the CM and simplify to get
\=\v?[q+q(j?h),xh,h\+\Fx(xk,h), A p ?\\v%[q+q{mh),xh,h] G>[q-q{mh)\ 2p\ + 1 ,
A=0,1 h= 0,1
(11)
(12)
= (j^Z
-vH[q+
?)
[G[q-q(j?x)] -G[q-q(m0)])
-
+ \(l-xi+x0)+j{vH[q+q(mxlxl,l]
q(m0), x0, 0] + F(x?,
?
1)
F(x0,
0)}.
(13)
846
G. Rocheteau,
P Rupert and R. Wright m to get l)m0 M, clears (Walras'
Next,
integrate
(6) over agents with diff?rent ?xx (1 p[?l)x0] =lr?i+(l-
which
says the money market clears law). Goods market clearing implies 75
iff the goods market *0
1+Xo?X\'
(14)
Also,
by Lemma
2:
?mx+(\-?)m0
={l[G(q)-G(q-qi)]
+
and money market clearing
(\-l)[G(q)-G(q-qo)]}^
implies =_???_
l[G(q)-G(q-qi)]+(l-l)[G(q)-G(q-qo)Y
one can proceed more generally, we focus on steady states Although are constant.11 infla where all real variables Then (15) pins down we can use the tion by p/p = M/M=l+j. Fisher Also, equation 1+ i= p/p? = (1 + 7)//3 to define the nominal interest rate /.12 Then (11>
(13) become \=\v?(q+qh,xh,h)+\Fx(xk,h), 1 /= 2 0 = (i+i)

+
A= 0. 1 A=0,1 (17)
(16)
G'(q-qh)
[G(q-q0-
G(q -q0)] +A(1
xx +x0)
11[vH(q+qi,xu\)-vH(q+qo,xo,0)+F(xi,?)-F(xo,0)],
(18)
where tions qi, for q(nth). System (16)?(18) in (x\, xo, q\, qo, A). Given this, aggregate we write constitutes five equa is given employment
by (14), the price level by (15), and individual money holdings by mh =p(l+7)[G(q) G(q-qh)]/?\.
11 are discussed for in a footnote below. See e.g. Lagos and Wright (2003) briefly Dynamics a general treatment of dynamics in these kinds of models. 12 as a no-arbitrage can be interpreted for pricing a hypothetical condition The Fisher equation off in the next CM, which asset purchased in one CM and paying nominal by assumption cannot defining ? to (traded in) the DM; or one can interpret itmerely be brought rate 7. i in terms of ? and the exogenous money growth as a piece of notation
Inflation
and unemployment
in general
equilibrium
847
A rium,
solution subject
to the above to one
caveat: we
set of equations defines a steady-state need to discuss the maintained
equilib assump
tion 0 < ?(m) < 1 for all m in the support of the equilibrium distribution
across agents entering the CM, upon which much of is the based.13 We now discuss conditions under which analysis is that is maintained valid. recall First, l(m) assumption given by (6), the ? ? can of which, p(\ + xo denominator be + mo m\, x\) positive or nega tive. We concentrate on the former case, in which we have ?\m) < 0, and to leave the latter as an exercise.14 After using money market clearing eliminate M and goods market clearing to eliminate ?, we have of money the above holdings ?( }- (l+7)(l-*i+*o)P*o + 0 -X1+XQ + 7*o)"*o -7*0^1 - (1 -xi+x0)(\+j)m (1+7)(1 -x\ +x0)[p(\ -x\ +*o) +mo-m\]
(19)
We use this to check whether 0 < ?(m) < 1 in equilibrium. Note that all agents who were buyers in the previous DM enter the CM with m = 0, while those who were sellers enter with the money they brought themselves plus what they acquired from sales. Letting m = max{m0, mi}, the most an agent could have entering the CM is therefore 2m. Since we are considering the case where ?'(ni) < 0, we have 0 < ?(ni) < 1 for all m in
the support of the equilibrium distribution iff ?(0) < 1 and ?(2m) > 0. By
(19) these
X\
inequalities
< 1
are equivalent
to: ~7(1 -*i)m
(1 -xi+x0
+ 7-7Xi)mi
/?(1+7)(1 -xi+xo)
7x0mi
Xo >-,
-(1
?x\ p(\
+Xo + 7-x;o)mo-2(l -+7)0 -x\ +*o)
~x\
+*o)^
can be checked for any solution to the set of equations defining which In some cases this is really quite easy. For instance, if utility equilibrium. = m are all is separable between q and (x,/z), so that m\?mo equal to
It is not that there is anything wrong in principle with equilibria with l(m) = 0 or l(m) = 1 for some m, but in practice reason is that, when the algebra becomes The main complicated. some agents hit corner solutions, we cannot guarantee all bring the same amount they will on employment of money then has to keep track of the (contingent status) to the DM. One DM money distribution one can use numerical closely 14 The related models. case where it is 0 is ruled out that the denominator is positive the more methods, as a state variable, which as e.g. Molico makes (2006) analytic or Chiu results difficult. and Molico (2007) However, do in
13
Note
in footnote condition mentioned 8. by the regularity iff the unemployed are better off than the employed in common the CM, which is actually if not the only possible if utility is case?e.g. h and (x, q), then the employed and unemployed separable between get the same consumption, while the latter enjoy leisure and hence are clearly better off. ? The editors of the Scandinavian Journal of Economics 2008.
848
G. Rocheteau,
P. Rupert
and R. Wright
M(l + 7), these inequalities reduce ratherdramatically to M M t

x\<\-and jco> ?.
can also express this in terms of consumption goods by insert ? ? = If, for example, utility is also p(\ + 7) [G(q) ing M/p G(q qh)] /?\. separable between x and h then things are especially neat, since then x and ? are independent of q, and since q < q* in any equilibrium, where q* is the first-best quantity, we simply need to choose preferences so that q* One is small. Intuitively, if the value of money is too big, either agents with m = 0 would have to supply i(m) > 1 to get back up to the equilibrium m = M(l+7), or those with m=2M to would have to supply l(m)<0 hence we need q to be not too big. In nu get back down to M(\ +7); merical it was not hard to choose parameters to guarantee calculations, 0 < l(m) < 1 for all relevant m, or to choose other parameters where the constraint ?(m) g [0, 1] is binding. In what follows, we do not dwell on this, and simply take for granted 0 < l(m) < 1 in equilibrium. that conditions hold so as to guarantee
VI. The Phillips Curve

We now study the relation between unemployment and monetary policy, where policy here can be described in terms of either inflation 7 or the nominal interest /, by virtue of (1 4- /) = (1 + 7)//?. We separate the analysis on buyers' preferences. into three cases, depending In the first case, as in most of the related literature, utility is separable between the CM and DM allocations (x, h) and q; in the second case q interacts with x; and in to reduce the notation, for the final case q interacts with h. Also, mainly this exercise we assume sellers' utility is separable in all three arguments, = the cases are: F(x) + G(q) + H(h). Summarizing, vL(q,x,h)
(1) Case (i):vH(q,x,h) = f(x,h) + g(q) (2) Case (ii):vH(q,x,h) = f(q,x) + g(h) (3) Case (iii):vH(q,x,h) = f(q,h) + g(x).
In the first case, (17) reduces to
= Since (20) has a unique solution for qh, we have qi=q0 q. Moreover, we can solve for q independently of the rest of the equilibrium. Then (16)?(18) reduce to
\=\[fx(xh,h)
?
+ F'(xh)],
A= 0,1
(21)
Inflation
and unemployment
in general
equilibrium
849
0=\[f(xul)
These three equations
+ F(xl)-f(x0,0)-F(x0)]-\(xl~xo-l).
can be solved
(22)
for (xj, x0, A), from which we get and the rest of the equilibrium. Note that policy i im ?=xo/(\+xo-x\) on with but not at all on (jci, jtn, A) or ?. One might q, pacts dq/di<0, a as this of in the cur version the neoclassical dichotomy, which recognize rent context has the implication that the Phillips curve is vertical.15 1. In case (i), dqldi < 0 and dl/di First, =0. (16) and (17) now become
Proposition In case
(ii) things are quite different.
A=i[/x(?+?A,x?) fq(q+qh,Xh)
Gf(q-qh) These conditions ?= ? = \ [g(0)~ rium conveniently
+ Ffe)], l
A=0,
* = 0,1
1.
= and qi=q0 q. Then (18) pins down imply x\=xo=x ? can we H Now summarize (I)]. g(l)-\- H(0) equilib as the solution (q,x) to
0= fx(q+q,x) 0 = fq(q+q9x)-(2i
=
+ FXx)-2\
(23) (24)
-
+ l)G'(q-qh).
=
< 0 and dx/di Therefore, dq/di 2G'(fxx + F')/D ~fxq, -2G'fxq/D where a ^ b means a and b are equal in sign and D = + + + + \)G" > 0. fxxfqq (/? F;/) (2i f2xq F"fqq falls with /, while the effect on x depends on the Thus, q unambiguously and since increases iffx increases. ~??x employment cross-derivativefxq, Proposition
15 As subset supply
2. In case
(ii), dqldi < 0 and dl/di
> 0 iff < 0. fxq

is said with to dichotomize if a
(1979, p. 43) puts Sargent can determine of equations no playing values role of
model it, "A macroeconomic the values of all real variables
in determining the equilibrium value of any the level of the money the real variables, supply helps determine but cannot values of all nominal that are endogenous variables influence equilibrium In a system real variable. that dichotomizes the equilibrium of all real variables values equilibrium
the level of the money real variable. Given the the any are
into real and of the absolute here does not dichotomize independent price level." The model can only affect the latter, since q is a real variable; nominal it instead parts, where money dichotomizes under this special into CM and DM parts, just like the baseline specification see Aruoba and Wright and Wright Waller (2003). Aruoba, in the CM and used that is produced capital by introducing DM goods. The plan here is to break the dichotomy by considering non-separable between CM and DM goods. model; Lagos-Wright break the dichotomy ? (2005) to make utility
850
G. Rocheteau,
This is very intuitive. Inflation is a direct tax on reduces q. lffxq > 0 (q and x are complements) then and hence ?. But i?fxq < 0 (q and x are substitutes) x and ?. In the latter case, inflation causes people goods
and into CM goods, increasing ployment. We thus get a Phillips curve under simple and natural conditions. for Perhaps the most surprising part is that the results are so clean?why, are no rea there wealth and effects? The substitution example, ambiguous son is the same as the reason why the model is so tractable, in general: agents here act as if they have quasi-linear utility. Before moving that this model has a neat graphical on, we mention = and Condition (11) implies x=X(q) (23) implies q representation. Q(x), which have slopes in (x,q) space that depend onfxq\
dq_ dx
activity, and hence inflation also reduces x, then inflation increases to substitute out of DM CM production and reducing unem
DM
= X(q) \x
-(/? + F"> and <"

"X\ fxq
xq -A
q= Q(x) (fxx+ G")
can be imposed to Figure 1 displays the three cases. Standard assumptions curves A these must existence. cross, guaranteeing imply simple calculation > one same curve must D that shows shows the be steeper than 0) (the X(q) the Q(x) curve, implying uniqueness of monetary steady state.16 An increase in / shifts the Q(x) curve down, leading to a fall in q and increase or a < 0 > 0. Hence, existence, uniqueness in x when we decrease orfxq have/^ in one simple diagram.17 and comparative statics can all be described case so we similar in sketch the analysis. (iii), merely happens Something becomes First, (16)
\=\[g,(xh)
which implies x\ =xo
+ F'(xh)], "I
A=0,1,
(25)
= x, but now
(17) becomes
1 \fg(q+qh,h)
with 17
is a value for x solving is of course also a non-monetary (23) steady state, which that here. 0, but we are ignoring can also depict in for 0<t(m)< the necessary and sufficient conditions 1, since < ?\ - [G(q) - G(q - q)], which we can this case they reduce to [G(q) q)] < ?\x G(q in the figure using Proposition easily plot in (x,q) space. One can also describe efficiency q One = to discuss Finally, we use the figure to show that if we peg the interest rate but if we peg 7, then in addition equilibrium; the non-monetary steady state over time along it is not standard methods: using dynamics i, then the steady state is the only monetary to there exist dynamic equilibria converging curve. the X(q)
16 There
4 below. hard
Inflation
and unemployment
in general
equilibrium
851
Q(x)
A/>0
I-1X*
(0/^=0
Fig. 1. Equilibrium representation
(ii)/^<0
(iii)/x,>0
which implies becomes
qo ^
q\
in this case, with
q\>qo
ifffqh>0.
Then
(18)
0 =?[f(q+qul)-f(q+qo,0) (I +i) [G(q-qx)We
+ H(l)-H(0)] G(q q0)] A.
(27)
~ or
of dqhldi < 0 for h = 0, 1 and dl/di leave as an exercise verification once increases The again, that employment important point is, ?fqh. decreases with inflation under simple conditions on the cross-derivative. Proposition 3. In case (Hi), dqhldi < 0 and dl/di > 0 iff < 0. fqh
inflation re This result is also very intuitive. As always, in the model, > or are 0 (q and h duces q. Iffqh q and leisure substitutes) complements, But if the reduction in q increases leisure and hence reduces employment. < 0 (q and leisure are complements) then, by reducing q, inflation also fqh reduces leisure and hence decreases Again, we equilibrium unemployment. or a curve Whether under very simple and natural conditions. get Phillips curve a an or a sees one not vertical upward-sloping, downward-sloping, in the data is another issue altogether, and not one we consider here. The for a relation point is simply that there is no problem in theory accounting inflation and between unemployment. Our results are all based on rudimentary public finance considerations, and do not require anything "tricky" like signal extraction problems, nom is that these con inal rigidities, and so on.18 The other point to emphasize siderations
18 One
lead to a relation between
inflation and employment
that is stable
to get unam is not used to compared to
set things up here. We say that there is one "tricky" ingredient might and labor that the DM is a pure exchange market, results by assuming as long as x production labor intensive, is relatively q. More produce generally, biguous ?
852
G. Rocheteau,
our in the long run. Thus, given the right cross-derivatives, and exploitable can achieve a permanently lower indicate that policymakers propositions rate of unemployment at a faster rate. However, we now by printing money argue that this is not a good idea. To this end, consider the planner's problem: W = max
(?,Xl,XO,?i,0o)
? [I
\-v"(q+quXu ^ ?
I
1)+ ?r? i 1? I
1-i
?
vn(q +<?o,*o, 0)
+ -vL(q-quXUl)+??s.t. ixi+(l -i)x0
vL(q-q0,x0,0) <?.
+ ?W
It is easy to check that, given our assumptions, the first-order conditions for iff / = 0. Hence, the this problem coincide with the equilibrium conditions efficient policy is the Friedman rule, regardless of whether the Phillips curve or vertical. Of course, we set up is upward-sloping, downward-sloping, the framework and so on. If to be difficult case in which it is although addressing following: Proposition
ner s problem.
avoiding externalities, market incompleteness, to add features such as these, it should not to be too high or too low, get equilibrium unemployment the analysis might lead to different policy conclusions? not so clear that inflation is the only or the best tax for these issues. In any case, in the benchmark model we have the intentionally one is free
4. The Friedman
rule
i= 0 achieves
the solution
to the plan
VII.
Conclusion
are desirable and can randomized allocations economies, unem as not This with lotteries. generates only supported equilibria it it is because for convenient very monetary provides theory, ployment, an alternative to Lagos and Wright in the sense that it generates (2005) tractability without quasi-linearity. We used these ideas to construct a gen In non-convex be We inflation and unemployment. eral equilibrium model of the relation between on prove that the Phillips curve slopes either up or down, depending of the utility function. The idea is simple and plausible: cross-derivatives inflation is a tax on economic activity that uses cash, and so if either labor intensive goods are substitutes for this activity or leisure is complementary This does not mean we with this activity, inflation reduces unemployment.
to get q, it will be possible content will be similar. ? similar the exact conditions will but the economic
results;
change,
Inflation use
and unemployment
in general
equilibrium
853
should
inflation
complications) In Berentsen
to combat unemployment, since (without additional the optimal policy is the Friedman rule. et al. (2007), a very different model of the labor market?
the one provided by Mortensen and Pissarides (1994)?is integrated into economics. There are some advantages to that search-based model monetary of unemployment, including the fact that it generates more interesting indi vidual labor market histories, and also more interesting aggregate dynamics, is a state variable (it takes time for individuals to because unemployment find jobs). On the other hand, as is well known, the Mortensen-Pissarides model is analytically one cannot tractable only with linear utility. Hence, to quasi-linearity for monetary say it provides an alternative economics, and one certainly could not attempt to prove theorems like the ones in this curve depends the slope of the Phillips on cross paper, about how of the utility function. But it is good to know that both of derivatives of the labor market can easily be generalized these models to accommo date monetary exchange.
References
N. (1991), Existence S. R. and Wallace, of Steady States with Positive Consumption Aiyagari, in the Kiyotaki-Wright Review Studies Model, 58, 901-916. of Economic G. and Puzzello, D. (2007), Markets and Monetary C, Camera, Aliprantis, Anonymous Economics 54, 1905-1928. of Monetary Trading, Journal S. B. and Wright, R. (2003), Search, Money, and Capital: Aruoba, Journal and Banking Credit, 35, 1085-1105. of Money, S. B., Rocheteau, G. and Waller, C. (2007), Bargaining, Aruoba, nomics 54, 2636-2655. Aruoba, S. B., Waller, C. and Wright, R. (2005), Money of Pennsylvania, manuscript, University Philadelphia. A. and Rocheteau, G. (2003), Money and Berentsen, Review Economic 44, 263-297. Berentsen, Run, A., Menzio, G. manuscript, University D. G. and Corbae, Camera, Review 40, 985-1008. A Neoclassical Journal Dichotomy, Eco
of Monetary
and Capital the Gains
and the Value from
of Money,
Trade,
International in the Long Economic Bank
R. (2007), and Wright, Inflation of Pennsylvania, Philadelphia. (1999), Money and Price and Welfare Current Real
and Unemployment International
Dispersion, Costs of
M. Chiu, J. and Molico, (2007), Liquidity of Canada, Ottawa. L. and Eichenbaum, M. Christiano, (1992), gate Labor Market Cooley, American Fluctuations, T. J. and Hansen, G. D. Economic Review Wage American
Inflation,
manuscript,
Business Review
Economic Inflation
(1989), The 79, 733-748.
Tax
and Aggre Cycle Theory 82, 430^450. in a Real Business Cycle Model,
R. (1987), Cooper, and Macroeconomic Corbae, change, Dutu,
D., Temzelides, Econometrica Julien, B.
and Employment in Labor Contracts: Patterns Microfoundations Harwood Academic NJ. Publishers, Newark, Applications, T. and Wright, R. (2003), Directed Ex and Monetary Matching 71, 731-756. Competing Auctions in a Monetary Economy,
R.,
manuscript,
University
I. (2007), and King, of Auckland.
854
G. Rocheteau,
P. Rupert and R. Wright in an Economy with Villages, manuscript, University of
Faig, M. (2005), Divisible Money

Toronto. Galenianos, M.
P. (2006), Multilateral and Monetary and Kircher, Matching Exchange, of Pennsylvania, University Philadelphia. and Lottery T. and Shell, K. (2004), Garratt, R., Keister, Comparing Equilibrium Sunspot Economic Review The Finite Case, International Allocations: 45, 351-386. Equilibrium manuscript, Green, Money Hansen, nomics Kiyotaki, Economy Kiyotaki, American E. J. and Zhou, R. (1998), A Rudimentary with Divisible Model Random-matching and Prices, Journal Theory 81, 252-271. of Economic Eco G. D. (1985), Indivisible Labor and the Business Cycle, Journal of Monetary 16, 309-327. N. and Wright, 97, 927-954. N. and Wright, Economic N. (1998), R. R. Review (1989), (1993), 83, On Money A 63-77. is Memory, Journal of Economic Labor Market Princeton Theory Fluctuations, Press, 81, 232 in T. F. Princeton, as a Medium of Exchange, Approach Journal of Political Economics,
Search-theoretic
to Monetary
Kocherlakota, 251. Kydland, Cooley NJ. Lagos, R. F. E.
Money
and Aggregate (1994), Business Cycles (ed.), Frontiers of Business Cycle Research, and Wright, R. (2003), Dynamics,
University
Dynamic, Fundamentally 109, 156-171. Lagos, R.
Disaggregative
in Genuinely and Sunspot Equilibria Cycles Economic of Money, Journal Models Theory of for Monetary Theory and Policy Journal of
Analysis, Lucas, R. Economic
R. (2005), A Unified Framework and Wright, Journal 113, 463-484. of Political Economy and Prescott, Theory 7, E. C. 188-209. (1974), Equilibrium Search
and Unemployment,
E. R. (1997), Journal Search Equilibrium. 105, Moen, Economy Competitive of Political 385^111. Interna in Search Equilibrium, and Prices M. of Money Molico, (2006), The Distribution tional Economic Review 46, 701-722. Mortensen, in the Theory of C. (1994), Job Creation and Job Destruction D. and Pissarides, Studies Review 61, 397?415. of Economic Unemployment, J. (2006), Aggregate Labor Supply: A Statement E. C, Rogerson, R. and Wallenius, Prescott, Arizona State University, and their Interaction, manuscript, about Preferences, Technology, G.
in Competitive R. (2005), Money in Search Equilibrium, and Wright, Equi Econometrica in Competitive Search Equilibrium, 73, 175-202. R. (2006), General with Non Rocheteau, G., Rupert, P., Shell, K. and Wright, Equilibrium of Cleveland, Bank and Money, convexities, paper, Federal Reserve Working Sunspots, OH. librium, and Rocheteau, in General R. (2007), Inflation and Unemployment P. and Wright, Rupert, Santa Barbara. of California, University manuscript, Eco and Equilibrium, Journal Indivisible Labor, Lotteries (1988), of Monetary 21, 3-16. in Economies with Efficient R. (1988), R. and Wright, Involuntary Unemployment G., Economics
Tempe. Rocheteau,
Equilibrium, R. Rogerson, nomics
Rogerson, Risk Sharing, Journal of Monetary Sargent, T. J. (1979), Macroeconomic
22, 501-515. Press. Theory, New York: Academic Indivisible L. (2006), Do Taxes Explain European Employment? Sargent, T. J. and Ljungqvist, and Savings, CEPR Discussion Lotteries, Labor, Human Paper no. 6196. Capital, Economic R. (1993), and Sunspot Equilibria, Lotteries, Indivisibilities, Shell, K. and Wright, Theory 3, 1-17.
Inflation
and unemployment
in general
equilibrium
855
Shi, S. (1995), Money and Prices: A Model of Search and Bargaining, Journal of Economic Theory 67, 467^96. Shi, S. (1997), A Divisible SearchModel of Fiat Money, Econometrica 65, 75-102.
Shimer, Stockman, R. (1996), Contracts NJ. Inflation Economics Search, and the Capital 8, 387-393. Money, International Balances Stock in a Cash-in-Advance Journal Review of Political 42, 847 in a Frictional Labor Market, Ph.D. Dissertation, Princeton University, Princeton,
A. (1981), Anticipated Journal Economy, of Monetary R. (1995), Trejos, A. and Wright, 103, 118-141. Economy Wallace, 869. N. (2001), Whither
Bargaining,
and Prices, Economic
Monetary
Economics?,
and Aggregate Real Individual Zhou, R. (1999), Economic Review International 40, 1009-1038. Existence of a Monetary Zhu, T. (2003), Steady
in a Random-matching in a Matching Model: Model: Divisible
Model, Indivisible
State
Journal 112, 307-324. Money, of Economic Theory of a Monetary Zhu, T. (2005), Existence Steady State Journal 123, 130-160. of Economic Theory
in a Matching
Money,

Wiley The Scandinavian Journal of Economics

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Wiley The Scandinavian Journal of Economics

Transféré par

Droits d'auteur :

Formats disponibles

Inflation and Unemployment in General Equilibrium Author(s): Guillaume Rocheteau, Peter Rupert and Randall Wright Source: The

Scand. J. of Economics 109(4), 837-855, DOI: 10.HH/j.1467-9442.2007.00511.x

Inflation and Unemployment

exchange, as in Lagos and money, between erence policy; Keywords'. JEL

it turns out, however, curve; ?40;

P. Rupert and R. Wright

tractability is provided by Shi (1997). See also Faig (2005).

The editors of the Scandinavian Journal of Economics 2008.

II. Basic Assumptions

The editors of the Scandinavian Journal of Economics 2008.

P. Rupert and R. Wright

The editors of the Scandinavian Journal of Economics 2008.

s.t. l{px\ where fer,

i)(pxo + mo) <wt

P. Rupert and R. Wright

for i e (0, 1) for now

yields i as the following function of m: pxp + mp-^M-m o o? \

? e (0, I) for Lemma 1. Assuming and W\m) = X/p are independent

to rule out p{x\ implies a singularity for one

even Actually, see Rocheteau

special in this case, our et al. (2006).

The editors of the Scandinavian Journal of Economics 2008.

+ ?W(msVmb) = G(q) + ?W(ms),

Inflation and unemployment

bargaining is the solution

\{vH[q+q(mJh),xh,h]+?W(0)} + \E?{vL [q-q(p),xh,h] + ?W(m+ p)},

since with seller.

1/2 he is a buyer probability Inserting the separable form of i/

V(mh,xh, h) = \{vH [q+q(mh\ xh, h] + ?W(0) } + lE,{F(xh,h) + G[q-q(p)] + ?W(n + p)}.

[q+q(mh), xh, h] + {Fx{xh, A).

\=\v?[q+q(j?h),xh,h\+\Fx(xk,h), A p ?\\v%[q+q{mh),xh,h] G>[q-q{mh)\ 2p\ + 1 ,

The editors of the Scandinavian Journal of Economics 2008.

P Rupert and R. Wright m to get l)m0 M, clears (Walras'

(6) over agents with diff?rent ?xx (1 p[?l)x0] =lr?i+(l-

iff the goods market *0

(13) become \=\v?(q+qh,xh,h)+\Fx(xk,h), 1 /= 2 0 = (i+i)

G(q -q0)] +A(1

The editors of the Scandinavian Journal of Economics 2008.

to the above to one

set of equations defines a steady-state need to discuss the maintained

to: ~7(1 -*i)m

+Xo + 7-x;o)mo-2(l -+7)0 -x\ +*o)

M(l + 7), these inequalities reduce ratherdramatically to M M t

VI. The Phillips Curve

The editors of the Scandinavian Journal of Economics 2008.

(ii) things are quite different.

(ii), dqldi < 0 and dl/di

> 0 iff < 0. fxq

model it, "A macroeconomic the values of all real variables

The editors of the Scandinavian Journal of Economics 2008.

P. Rupert and R. Wright

-(/? + F"> and <"

The editors of the Scandinavian Journal of Economics 2008.

which implies becomes

in this case, with

0 =?[f(q+qul)-f(q+qo,0) (I +i) [G(q-qx)We

+ H(l)-H(0)] G(q q0)] A.

lead to a relation between

inflation and employment

The editors of the Scandinavian Journal of Economics 2008.

P. Rupert and R. Wright

+ -vL(q-quXUl)+??s.t. ixi+(l -i)x0

The editors of the Scandinavian Journal of Economics 2008.