My ¬rst foray in this intriguing domain, inspired mostly by early work

of John Roberts (1989) and Jean-Jacques Herings (1996), dates back to

1997 [Drèze (1997)]. It is surveyed in my Presidential Address to the

International Economic Association [Drèze (2001a)]. A general model,

studied in a recent joint paper [Citanna et al. (2001)] is now being

extended with Jean-Jacques Herings to the incomplete markets frame-

work. This work illustrates the merits of the general equilibrium methodo-

logy for tackling macroeconomic issues. Although coordination failures

have come to the attention of macrotheorists along other routes”partial

equilibrium or macromodels, surveyed by Cooper and John (1988)”the

link to price rigidities emerged from the general equilibrium analysis.

Licandro: Once again, we face a technical issue that deserves some

explanation . . .

Drèze: Let me try. Consider ¬rst an economy consisting of one

¬rm turning labor, supplied by households, into output. Returns are

diminishing. Nominal wages are given. Households jointly supply N units

of labor, collect wages and pro¬ts, and buy output. The ¬rm maximizes

pro¬ts. Any level of output using no more than N units of labor de¬nes

an underemployment equilibrium, with output price equal to marginal

cost. Indeed, the ¬rm maximizes pro¬ts, households optimize under a

constraint on labor supply, and markets clear.

Dehez: That is at variance with the three-good model of Barro“

Grossman (Barro and Grossman, 1976) and Malinvaud (1977), where

classical equilibria are unique?

Drèze: With reference to that model, the equilibria just proposed

de¬ne the frontier between classical and Keynesian unemployment, a

locus indexed by output prices at given nominal wages. In the three-

good model, all nominal prices are ¬xed. I explain in my 1997 EER

paper how this implies a particular selection from the continuum associ-

ated with ¬‚exible output prices. Actually, the continuum is a general

property. Thus, consider an Arrow“Debreu economy with two sets of

commodities, say F commodities with ¬‚exible prices and R commodities

with downward rigid nominal prices. Quantity constraints are allowed

on supply alone; we are looking for a “supply-constrained equilibrium.”

When the rigidities bite, the R ¬xed nominal prices imply that R ’ 1

relative prices are given. But R markets are allowed to clear through sup-

ply constraints. There is thus one degree of freedom left. It corresponds

to either the overall ratio of the ¬‚exible prices to the rigid prices, or to

the overall extent of rationing for the commodities with rigid prices.

Walras™s Law links these two macroeconomic variables as per a Phillips

An Interview with Jacques Drèze 293

curve of sorts. There remains a single degree of freedom, corresponding

to the selection of a point on that Phillips curve. The question then

arises: How is an element from the continuum selected? Note that com-

petitive equilibria also exist in my economy, at nominal prices high enough.

My provisional conclusion is that the intertemporal equilibrium model

must be complemented with a speci¬cation of the short-run adjustment

process that links successive equilibria. That speci¬cation should embody

the sources of price stickiness; it should cover the transition from one

multivariate equilibrium to the next”possibly as per a tâtonnement or

nontâtonnement in prices and quantity constraints, of which I have

studied some examples [Drèze (1991, 1999)]; and it should perform the

selection of a speci¬c equilibrium from the equivalent of a Phillips curve,

especially when the latter is multidimensional.

Licandro: Why multidimensional?

Drèze: When we move from Arrow“Debreu to the more realistic speci-

¬cation of incomplete markets, the degree of indeterminacy may rise,

but that is really technical! Well, you asked for it. In the two-period

stock-market economy with S states and J assets, J less than S, we may

expect S ’ J + 1 degrees of freedom, that is, a set of equilibria of

dimension S ’ J + 1! I have not encountered such a Phillips curve in the

macroeconomic literature yet. Not surprising, given the limited popular-

ity of multiple equilibria. But general equilibrium theory aims for gener-

ality. If your premises entail multiple equilibria, you had better ¬nd out,

and face the consequences!

Dehez: Your answers to the last two questions refer to nominal rigidities.

Your own interest in money is rather recent, I think. How does it ¬t into

the broader picture?

Drèze: In joint work with Herakles Polemarchakis (2001) and then

also Gaetano Bloise [Bloise et al. (2005)], we use a consistent and

natural de¬nition of a monetary economy: Money balances are used for

transactions; they are supplied by banks, which lend them at nominal

interest rates set by themselves. Thus, we are considering “inside money,”

the only kind issued by central banks with balanced accounts. At com-

petitive equilibria under given nominal interest rates, there remains in

such a model indeterminacy of the overall price level. In a one-period

model, one would say that all relative prices are determined, but the

overall price level is arbitrary”a standard feature of the Arrow“Debreu

model. In a multiperiod model with certainty, the same property holds,

and the in¬‚ation rates relating the price levels at successive dates are

determined through a Fisher equation. That is the starting point from

which different authors proceed toward determinacy along different routes,

like feedback rules or the ¬scal theory.

294 Pierre Dehez and Omar Licandro

However, in the intertemporal model with uncertainty”that is, with

alternative states of the world at any date”there is further indeterminacy

to the following extent: At any date event, the expected rate of in¬‚ation

between today and tomorrow is pinned down by interest rates, but the

variability of in¬‚ation rates across alternative realizations tomorrow is

unrestricted. Understandably, a single instrument, namely, the nominal

interest rate, implies a single constraint, at each date event.

Licandro: Of course, price-level determinacy in monetary economies

is a debated issue. Exactly what is your stance?

Drèze: The extent of indeterminacy just stated is both a headache and

a blessing: a headache because we all know that price levels do not jump

around like puppets, but also a blessing because indeterminacy in the

abstract model leaves room for endogenous nominal rigidities to pin

down price levels.

Licandro: Another enigmatic assertion! Can you explain?

Drèze: I started my Baf¬ Lecture at Banca d™Italia (1992) with the

question: “When warfare in the Gulf bids up oil prices, do you expect the

prices of books or magazines to go down?” Because many prices are set

at intervals, and because many are downward rigid, the answer is clear.

As relative prices vary, price stickiness generates some core in¬‚ation. This

is part of the short-run adjustment process selecting an equilibrium from

my continuum, from my Phillips curve if you wish.

Many macromodels of the New Keynesian vein go that route”through

staggered prices, menu costs, and the like. I differ on two scores: the

explanations of price stickiness”we talked about that, and the formal

analysis of its implications”which brings us back to the real and nominal

indeterminacy associated with price rigidities.

Licandro: And the upshot for macroeconomics is . . .

Drèze: The upshot is both substantive and methodological. On the

substantive side, I feel that coordination failures associated with price“

wage rigidities”whatever the origins of these rigidities may be”have

their place in macroeconomic theory and policy. I only wish that I could

measure the extent of coordination failures empirically, but that lies

probably beyond my own horizon. On the methodological side, I am

now investigating some macroeconomic implications of microeconomics

in a general equilibrium model extended simultaneously to incomplete

markets, money, price and wage stickiness, but also increasing returns

[Dehez and Drèze (1988)] and imperfect competition [Dehez et al.

(2003)]. The works! And a long way from the competitive Arrow“Debreu

model. . . . As we discussed, these extensions lead to multiple equilibria,

and the static intertemporal model remains to be complemented by a

speci¬cation of short-run adjustments, for which generality is an open

challenge.

An Interview with Jacques Drèze 295

Dehez: What equilibrium concept, or concepts, are you using?

Drèze: There is always a trade-off between generality, hence scope for

realism, and tractability! In models with arbitrary ¬nite horizons, which

are also the basic tool for in¬nite-horizon analysis, the perfect foresight

equilibrium of Radner (1972) is the easier starting point. It lends itself

well to my extensions, and to the analysis of coordination failures, but

perfect foresight is a strong assumption, especially under multiple equilibria,

and I want to pursue more general formulations in the spirit of “tem-

porary equilibrium” à la Grandmont (1977). Arbitrary horizons then create

logical as well as technical dif¬culties, with which I am currently struggling.

Licandro: All that seems quite remote from contemporary

macroeconomics!

Drèze: Indeed. My vision is that the extended model is susceptible of

encompassing macroeconomics! What I mean is: Most of the models

used in macroeconomics concern economies that ¬t within the general