The Analysis of Exchange Rate Movements: Neutrality Results and the Role of Expectations

SROs Cena eerie wr von OTD DNV CUT TET NOOIT VON DNOINT SOET UDO TER SY SORDID ODT EET T SOE nee

nalts anima tittle lt sie

an dade ek a ns an eR a a lca A Di eee ee ee oe ee

October 1975 |

The Analysis of Exchange Rate Movements: Neutrality Results and the Role of Expectations

.

by

=

Peter Isard and Michael G. Porter

.

NOTE: International Finance Discussion Papers are preliminary materials circulated to stimulate discussion and critical. comment, References in publications to International Finance Discussion Papers (other than an acknowledgement by a writer that he has had access to unpublished material) should be cleared with the author or authors. ;

setts nal Aca ta ataamserte itemise, mt tart maces mate toner sr minonmnnite science cin arene

: ; : : Peter Isard and . Hidhacl G. Porter* : . . October 1975

_The Analysis of Exchange Rate Movements: Neutrality Results and the

Role of Expectations

: . Following a bricf discussion of existing literature on the analysis of exchange rate movements (Section 1), this paper re-examines the conditions under which exchange rate movements have no effects on the real balance of trade in particular, and on the complete list of

real economic variables in general (Section 2}. Although these

neutrality conditions do not hold in the world as we know it today,

.they do point to. the types of institutional changes that would weaken

’

the real impacts of devaluations over time (Section 3).

t

When the neutrality conditions do not hold, ex-ante exchange rate expectations have an important influence on the timing and magnitudes of the effects of devaluations, The correctness of exchange rate expectations can be quantified for use in empirical studies of the effects of exchange rate movements (Section 4).

When the role of anticipation is introdsced, devaluations and revaluations are seen to share a common properzty with all other anticipated interventions in the private market, viz, they distribute income and wealth fror the many -to the few-- those with superior information receive capital gains at the (ultimate) expense of the rest of the community. These wealth effects. may conceivably dwarf the contribution */ Board of Governors of the Federal. Reserve System and Australian National University, respectively. The views cxpressed herein are solely

those of the authors and do not necessarily represent the views of the Federal Reserve System.

of other wealth effects to equilibrium outcomes, which suggests that the monetary approach to balance of payments theory needs

augmentation to allow for the role of anticipation (Section 5), 1. A Brief Discussion of Existing Literature

The textbook analysis of discretionary exchange rate changes is typically conducted within a two-region partial cquilibrium model of the trade balance, focusing on price elasticities and emphasizing

the Marshall-Lerner conditions, Researchers at the International

Monetary Fund have extended this elasticities appreach to model and

.-

simulate the effects of exchange rate realignments within a multi-

, 1 regional, multi-commodity framework,™

In xecent years the conventional elasticities approach has encountered several challenges. A monetary model of the balance of payments has been extended by Dornbusch (1973) and others, who argue that the primary effects of devaluations derive frozn attendant changes in the real value of money or wealth. “A separate but less formal challenge has been issued by Laffer (1974), who argues that nominal price adjustments following a devaluation will neutralize any impact on the real trade balance. These criticisms are in the tradition of much earlier arguments (see Meade and Russell, 1957) that the change in factor shares following a devalugtion would be unacceptable and ° would be follow ved by wage increases designed to neutralize the real

impact of devaluation.

-

i1/ Armington has been a major contributor to this research, but see Artus and Rhomberg (1973) for a recent description,

tlm n nln st eine na at cane te a

2 1 H i s

Pata eee

ote nee TA ham ment Me oa Sa ne ha SR rm tlhe a te renee

-~3-

In an unpublished paper, Henderson (1974) has further extended the analysis of devaluation to a. two- country, short-run macroeconomic model with one good, two interest-bearing paper assets (securities) and two currencies, It is shown that when monetary authorities refrain from monetary expansion ‘via open market sales of domestic securities following a devaluation, necessary and sufficient conditions for devaluation to leave the real trade balance unaffected are (1) perfectly flexible wages (full employment), (2) the absence of wealth effects on‘consumption, and (3) perfect substitutability betveen

the securities issued inthe two countries.—

2. A Generalized Statement of the Neutrality Conditions

In this section we present a’ generalized Statement of the conditions under which a devaluation is neutral in a multi-country, ‘Longrun (or multi-period), multi-commodity, multi- ~asset world, We have in

mind a modification and simple monetization of the abstract world of

Arrow and Debreu (1954)3/ | Our argument rests on the claim that in an

2/ The sufficiency of these three conditions is “suggested by the

following arguments, With perfectly flexible wages (full employment)

and one good, labor's equilibrium real income in each country is unaffected by an exchange rate change; and the absence of wealth effects on con~ sumption then precludes any change in wal consumer @emands. With perfect substitutability between securitics, exchange rate movements cannot cause equilibrium real interest rates to dive erge internationally: there are no market forces to push investment demands in opposite directions in the two countries, Hence, there are no forces to push absorption in opposite directions in the two countries, so markets clear with no changes at all

in real absorptions or the trade balance,

3/ The modification involves (1) a grouping ef consumers and producers by country, (2) a grouping of commodities distinguished by time period,’ and (3) the introduction of marketable assets which yield real endowment (income) streans over time to their owners, payable by debtors, ‘The first two alterations are trivial. The third alteraticn can be introduced without changing the topological properties (closedness and convexity) of the sets of feasible production plans and consumption vectors, and hence does not change the qualitative nature of the conditions that. are sufficient for the existence of a competitive equilibriun, |

ak ale a ctl te Ni hd Mable Aen

.” hw

m-country world with m-1 independent exchange rates in each of n-periods, if (1) all quantity-demand and quantity-supply fetions _- including asset supply functions -- are homogeneous of degree zero in prices, if (2) competition eliminates the profitability of commodity arbitrage internationally, so that (ignoring tariffs, ‘and transport costs) each

commodity has equivalent common-currency prices im all countries, and

if (3) a unique multi-period real equilibrium exists, then (4) the vector of prices, wages, interest rates and exchange rates consistent with real

equilibrium has m-1 degrees of freedom for each time period, or one degree

1 4 . of freedom for each independent exchange rate tt! Thus, we can conceive of

é --

4/ The proof of this claim can be sketched verbally. Start with a closed econony. For the single-period case, it is well-known that the equilibrium price vector has exactly one degree of -freedoin -- the absolute price level is indeterminant. For the two-period case, the equilibrium vector of prices and interest rate has exactly two degrees of freedom -- the absolute price level in each period is indatcrminant ~- provided money supplies and other exogenous nominal variables are denominated in realor purchasing-power terms (i.e., homogenous of degree zero in prices) and the nominal interest rate reflects the real rate of interest plus the rate of change in the absolute price level. For the n-pariod case, the equilibrium vector of prices and interest rates has exactly n degrees of freedom, . . a

Now consider the n-period case with m countries. Under autarky, each country has an cquilibrium vector of prices and interest rates with exactly n degrees of freedom. The movement to free trade adds (m-1)n independent exchange rates to the list of price variabies to be determined, but also changes the number of constraints on the gquilibrium price vector, Eech country now has.a balance of payments identity in each , period -~ a net addition of mn constraints. (Je are thinking in terms of a flexible exchange rate world, but official intervention is admissible as long as intervention functions are homogenous of degree zero in prices.) For each tradeable good, the opcning up of free trade eliminates the constraint that absorption equal prodductéon in each period (we are ignoring commodity storage over -time, which complicates the argument but docs not change the conclusions), while adding the constraint that absorption equal production for the world as a whole in each period -- a net elimination of (m-l)n T constraints, where T is the number of tradeable goods; (for simplification, we are assuming the sawe T in cach period), The climination of these (m-1) n T constraints is exactly offset by the addition of the constraints requiring in each period that the prices of cach tradcable good in countrics 2,.<., m equal the price of the sane yood in country 1 multiplied by the prices of currency 1 in tenn Of currencies 2, .445Me Thus there are mn additional ‘conditions and (m-1)n additional variables, or a loss of n degrees of freedom as compared to the autarky case, Leaving

we

n(in-1) degrees of freedom altogether.

NR ERM ETS RE ES ee Re Re RE RE

= 5 -

an abstract world in’which all. real variables in the economy move

me ae ¢

along their unique equilibrium time paths independently of the time . paths of exchange rates. ‘Provided that all prices, wages and interest J. rates were perfectly flexible and conditions (1) - (3) were met, fluctuations in the time paths of exchange rates would be neutral (in the sense of having no effects on any real econemic variables), and the nature of the exchange rate regime (fixed or flexible) would be a matter of indifference.

The conditions sufficient for neutrality here are analogous

to Henderson's. Our conditions (2) and (3) are also assumed in the

o

framework of Henderson's model, though not stated explicitly in his

theorem, In our framework perfectly flexible prices, wages and interest

rates, analogous to Henderson's perfectly flexible wages (full employment), are sufficicnt to clear all markets at the equilibrium relative price levels, independently of the time paths of cxchange rates.

: Our remaining condition of hemozeneity is: indirectly analogous to Henderson's conditions of perfectly swhstitutable securities and the absence of wealth cffects, Zero degree homogeneity in prices of all quantity-demand and quantity~supply functioms is a condition

which is violated in reality to the extent that quamtity~demands and

a ema a a cleaned hh chaaeariand Bir ndt dtm patna tema Bi en .

quantity-supplies depend on real wealths, and to the extent that real . wealths are not homogeneous of degree zero in prices, This homogeneity assumption seems reasonable only in a (fantasy) world in which money

supplies and all other forms of debt: are denominateed in real terms or

tee cnet mene RR RR ST Re Sena RRR ET CSR CR OS Se RTE SE Seep DMR a De SER SERN PE RT SO RR RE ENON LIE 9 LTE OS SCR TERR: WT" >

nanan ts A rasta en tin asl ence haben sie enahinamty satin +

Se acca a nee tee Le Aa ee ate omen cae en Rene ee te SIRES i et ree ee ee Ra ni a

. ry

purchasing power units, or one in which the government manages to

expand the nominal money supply at precisely the rate of increase of prices. In such a world, real wealths are liomogeneous of degree zero in prices, Henderson's conditions then tolé: devaluations neither

have wealth effects on consumption nor cause amy divergence between

real rates of interest on different assets,

We have suggested that apart from conditions which most

economists would probably accept as reasonable, sufficient conditions

for the neutrality of devaluation in a multi-country, multi-period, multi-commodity, multi~asset world are (a) perfectly flexible prices, wages and interest rates, with (b) money supplies and all other forms of debt denominated ja real terms. This conclusion extends from comparative static to dynamic models, Suppose, for example, that all debt was denominated in purchasing power units, and that relative prices were fixed (not necessarily at market-clearing levels) independently of exchange yates, through the indexing of commnedity prices, wages, and interest rates, In such a world the dynamic

paths of real variables would be unaffected by exchange rate movements (whether or not these movements vere correctly foreseen), since the ultimate determinants of all real variables -~ namely, relative prices and real wealths -- would be completely insulated from exchange rate mevenents, Such a world would insure not only the real values of financial assets, but also the real income streams of labor and physical capital and hence the real capitalized values of human and

physical assets,

STE RNR EL EM EE PONE IOI ORTOP ES SE TE ENC IE IT TETRA HS pe RA SIE EY ART TMI Te I" REE TR RSIS NA RES ORR ENT

_ .

‘Needless to say, the conditions sufficfent for the neutrality ‘of devaluation do not apply to the world as we know it €oday. In reality, . . money supplies and most other forms of debt are not denominated in real terms, and devaluations have impacts on xeal wealth, as we discuss in Section 5 below. Moreover, many prices, wagrs and interest rates are sticky, and discretionary exchange rate adips-tment has been viewed as a technique for adjusting relative prices across countries faster than would otherwise occur through market processes, (Floating the exchange rate can be viewed as a technique for making this adjustment

; automatic.) The stickness of many interest rates is due to legal

o

constraint, Wages are sticky because“suppliers of labor, lacking

.

perfect information and facing search and relocation costs, find themselves locked into firms and geographic regions. Many other prices are sticky because price-setting obligopolists have healthy profit margins to begin with, lack complete information about industry demand functions and the reaction functions of other oligopolists, and find it costly to. reassess pricing strategies and advertise new prices as frequent intervals,

Having argued that the conditions sufficient for the neutrality of devaluation do not hold in the world as we know it today, we may nevertheless note tendencics in today's world toward institutional

changes which may weaken the effectiveness of devalluations. Spec‘ ically,

devaluations may become progressively more neutral ever time with

institutional changes toward market adjustinent of prices currently the

Re ee ee ee a 2 AOE lS es leo

~8-

subject of intervention, toward a more rational €or real) denomination of debt, toward a reduction in the immobility of Labor and capital

or an indexing of nominal factor payments, and consistently, toward greater competition (less price setting and more price taking) or more price indexing in product markets,

4. The Role of Expectations

In discussing the conditions under which a devaluation would be neutral, we have been suggesting without much sophistication that

¢

: the time paths of real economic variables depend wltimately on relative i £ . ee.

prices and real wealths, rather than, more corréetly, on ex ante

. - 5/ ; expectations of relative prices and real-wealths.— Implicitly our

a a ee De Se

arguments have assumed that the conditions under which relative prices and real wealths are independent of exchange rates are also conditions under which ex ante expectations of relative prices and real wealths are independent of exchange rates, This assumption is consistent with the hypothesis of rational expectations.

For a world in which the neutrality conditions do not hold, however, ex ante expectations of relative prices and real wealths will indeed be quite sensitive,to exchenge rates, or more precisely, to exchange rate expectations, Provided that cconomic participants are able to act on the basis of their expectations, the effects ofa devaluation may be quite sensitive to the degree to which it is anticipated,

>/ In addition to ex ante expectations held currently, past errors in

ex ante expectations can lead to behavioral adjustment, thereby affecting current real economic variables,

cca Asin ams acne ciate inal Ney ade te wan meme

pectations, .

-~ 9.

(4: ue" .° This suggests that empirical work on the timing and magnitude of the effects of exchange rate movements should take into

account the nature of exchange rate expectations, Three important

- properties of expectations have been discussed by Despres (1967):

.(i) the degree to which they prove correct éx post; (ii) the degree

to which they zre held uniformly by all economic purticipants;

and (iii) the degree of precision or confidenee with which they are held, The first property, correctness, is relevant: to whether actions taken as‘a result of expectations prove to be optimal

reactions ex post. The ‘second property, uniformity, is relevent to the

’

‘distribution effects of an expected or unexpected event, insofar as

some economic participants may be caught more by surprise than others.

.

The third property, precision, is relevant to the ections that risk-

averse participants will choose to take on the basis of their ex-

- The uniformity and precision of expectations are difficult to quantify. The correctness of ex-ante exchange rate expectations at various distances (i.e., over various time intervals) can be quantitifed, however, Specifically, we can quantify errors in thie exchange rate . expectations. held at various distances from ex post data on actual exchange rates, together with computations of the expected future spot

rates consistent with ex-ante data on either forward exchange rates

is : . . . 6 . or international interest rate difforentiais £/ Giwen the current spot

/ See Porter (1971).

cea a ne Aen ae cena a nie ae te Gree Te oe em eee

fe

Aernteatnaa atin stile tatads cle ssnktbest bth Ste Rat. st te seer

“rate by a percentage equal to cither the premium on forward exchange

- 10 -.

exchange rate, the spot rate expected to prevail on any specific date

in the future can be quantified as differing from the current spot

for delivery on the specified future data, or the difference in international interest rates attached to comparable securities maturing on

the specified future date,2/ If we could hold constant all other

factors governing the response of wealth and prices to devaluation,

we might expect that the responses, or changes. o2 trade volumes and & P I > Bes,

other real variables would (in absolute value) be increasing functions

of the error -in exchange rate expectations at any given distance,

Although we do not propose, in this paper, to fit expectations

ot

erro into an empirical model of the trade balance or the overall

balance of payments (under an adjustable peg regime), it seems clear that both elasticities approaches and monetarist analyses can mislead . . wa , 8/ .,.

if they fail to consider expectations errors. With respect to

i/ It may be noted that small fluctuations over time in international interest differentials on a given maturity (e.g., 90 day bills) are consistent with large fluctuations over time in expected future exchange rates if spot exchange rates exhibit large fluctuations, Thus, perfect knowledge of the term structure of international Enterest rate differentials is consistent with highly imprecise expectations of future exchange rate levels; spot exchange rates may move substantially over short periods in performing their function of maintaining market clearance,

8/ Sophisticated macroeconomic models in use teday do take expectations into explicit account, however, and have thereby initiated an important focus on expectations errors, See Shafer (1975) for exainple, who develops a simulation model in which real economic behavior depends on expectations of variables such as price levels, the ratios of market value to replacement cost for physical capital, consumer income and wealth, and exchange rates

De

°

FIN eR RE ena Oe ee oT ERENCE A NRO FIRE TTR A OE EE RITE AE TNO SN NEI IS the cen ME IEE 2 OT OL ETI I TE STINE NREL

iti aecae hence tac Menai duit thi iain ae ile nl hae on ae te

- 1) -.

s elasticities approaches, the timing of the first responsesof

suppliers and demanders to a devaluation depends on how their

‘expectations develop; and although the adjustment of prices in response

to expectations can in theory remove incentives two change any real variables, the presumption in fact is that both the timing and scale or real adjustments are influenced by expectations.

5. Anticipation and the Sive of Wealth Iffects

Monctarist analyses of devaluation must also be aware of

pitfalls, To illustrate we consider the model of Dornsbusch (1973),

who argues that devaluation has a positive real wealth effect on foreigners

(i.e. holders of the appreciated currency), and anegative real wealth \ PP y); &

’

effect on donestic residents (i.e., holders of the: depreciated currency), which in the short rua increases foreign spending relative to production while reducing domestic absorption relative to production. Thus, in the short run the wealth effects of devaluation serve to unambiguously improve the trade balance of the devaluing country.

But what if we modify this model to allesr for anticipation? It is argued here that correct anticipation of any: change in any official and adjustable peg (exchange or interest rate) will result in an increase in private’ wealth at the expense of the public . sector, The expectation of revaluation induces the private sector to incur increased foreign liabilities, selling the foreigm currency to the central bank until expectations have been validated, If the stock of

Pa

forcign liabilities incurred in anticipation of a 100g per cent

see men I CR eT MI ES A mR I ER RE RENE CRIES RETR CO SONG EET RRR: FOIA RTE re

ead I a RR ee nen ie a RE ae KE Fee ma RR ae ies ei ll pla Tilsen Natacha nat: i. acme

~ 12 -

revaluation is X, private holdings of base money will have increased

.by gX units as a result of actions’ taken in response to the expectation

of revaluation, Whether there is a corresponding rise in real wealth of gX depends on whether the matching losses of the central bank are viewed as leading to a higher real tax burden. In the limiting case, where the government raises taxes to cover the losses of the central bank, and where taxpayers expect this, then anticipation of exchange rate changes does not involve a wealth effect. As an alternative case, the private sector may consider itself liable for paying off privatelyheld government bonds without simultaneously considering itself liable for high-powered money debt, In this case, whether or not a wealth effect result’ from the injection of the gx units of base money depends, from an alternative viewpoint, on whether taxpayers expect the stock of base money to remain permanently higher than it would have been in the absence of the revaluation, It should also be noted that Dornbusch's wealth_effects would disappear if the private sector considered itself liable for both privately-held government bonds and high-powered money debt. |

Anticipation of devaluation leads to similar results, except the ability to obtain foreign assets prior to devaluation may be constrained by official reserves, But if we assume symmetry, then the same gX is again there for the asking .in the event of a fully anticipated devaluation, Whether the stock of base money actually rises by

this amount (gX) depends upon whether the foreign assets are

manana, ila ee ctl dine RN eh le

(Nn ee Amey mI ee

~13- oc:

e

reconverted into domestic currency after the devaluation. All we can

say, in general, is that any correctly anticipated change in an official

price, i.e. a price that is the subject of intervention, leads to an

increase in real ‘private wealth (abstracting from expected tax offsets),

‘some proportion of which is.in the form of an injection of base money.

=. The increase in private wealth that results from correctly anticipating exchange rate changes will be distributed among foreigners and nationals according to their relative degrees of foresight and ability to move funds across currencies, An increase in domestic wealth

is likely to spill over into increased domestic gemand, which would tend

to worsen the domestic trade balance, whereas an increase in foreign

Py ’

wealth would have the reverse tendency. For the case in which the

domestic wealth effect from anticipating devaluation dominated the foreign wealth effect, it is conceivable that the direct "Dornbusch" effect would be offset, -With an anticipated revaluation, on the other hand, the direct wealth effect would be reinforced by the increased private wealth flowing from correct anticipation. ‘The reverse is true

when the foreign wealth effect from correct anticipztion dominates the

-domestic wealth effect, Thus we find that where expectations are

correct, the monetarist model may prove wrong in suggesting that a

devaluation improves the balance of trade in the short run. A stronger

conclusion is that empirical attempts to estimate the magnitudes of the wealth cffects of devaluations can be seriously misled if they fail to

capture the degree of anticipation,

Fer ea aOR

The first several sections of this paper suggest that

. devaluations would be neutral in a world in which moncy supplies and

all other forms of debt were denominated in purchasing power units,

and in which prices, wages and interest rates were perfectly flexible, Although these conditions do not apply to the world as we know it today, they do point to the types of institutional changes that could be expected to reduce the real impacts of exchange rate changes -- namely,

institutional changes toward market adjustment of prices currently

the subject of intervention, toward a more rational: (or real) denomination

of debt, toward a reduction in the immobility of labor and capital or

an increase: (formal or informal) in the indexing of nominal factor

payments, or toward greatcr compctition (less price sotting and more price taking) or an increase in price indexing in product markets. When the neutrality conditions do not hold, the nature of ex ante exchange rate expectations has an important influence on the timing and magnitudes of the effects of devaluations, Three properties of exchange rate expectations can be distinguished: correctness, uniformity, and precision, Of these, the correctness of exchange rate expectations can be quantified and should be used as an explanatory variable in cnpirical studies of the effects of exchange rate movements. We have also argued that the wealth effects which dominate

monetarist analyses of devaluation may be quite sensitive to the extent

to which the devaluation is anticipated, In the case of strong and

ened eee aa ate heretic nee len meet ie fn Anna enn cnoe i

vem tee

ee lB Sh sek

correct anticipations it is conceivable that the wonectarist model may prove wrong in suggesting that a devaluation improves the balance

of trade in the short run.

~~

Aen ella Nell le te lh ha celal Mla eh it oct ey eee!

Arrow, Kenneth J, and Gerard Debreu, "Existence o£ an Equilibrium fora Competitive Economy", Econometrica, July 1954.

Artus, Jacques R, and Rudolf R. Rhomberg, "A NulitiLateral Exchange Rate Model", I.UM.F, Staff Papers, Novanber 1973.

Despres, Fmile,. “Inflation and Development: Brazit™, transcript of a talk given on May 1, 1967, reprinted im Gerald M. Meier, International Economic Reform: Coltectel Papers of Emile

Despres, New York: Oxford University ress, 1973,

Dornbusch, Rudiger,"Currency Depreciation, Hoarding and Relative -Prices",’ Journal of Political Economy, July/August, 1973,

Henderson, Dale W., “Monetary Fiscal, and Exchange Rate Policy in a Two Country, Short Run, Macrocconomic Mode 1", revised manuscript, December 1974,

Laffer, Arthur-B., “the Bitter Fruits of Devaluation", Wall Street Journal, January 10, 1974,

Meade, .J.E, ‘and E, A. Rus sell, "Wage Rates, the Cost of Living and the Balance ot Payments," Economic Record, April 1957.

Porter, Michael G., "A Theoretical and Empirical Framework for Analyzing the Term Structure of Exchange Rate Expectations", I.M.F. Staff Papers, November 1971.

Shafer, Jeffrey R., "Monetary Policy Under Floating Exchange Rates: Simulation: of a Two-Country Model of Macroeconomic Interaction, ! manuscript, July 10, 1975,

eR oe re aa en RS eR TREE EL IRN REE OTE TERRI EN POAT EAI tM SERV Oe OF Si AR REAP 6° SN OCT, ERE A IIE NTT