Download Understanding Hyperinflation: Patterns, Dynamics, and Links to Budget and Money Growth and more Study notes Dynamics in PDF only on Docsity!
RUDIGER DORNBUSCH
Massachusetts Institute of Technology
FEDERICO STURZENEGGER
Massachusetts Institute of Technology
HOLGER WOLF
Massachusetts Institute of Technology
Extreme Inflation: Dynamics
and Stabilization
WAR AND REVOLUTIONwere once the settings in which extreme inflation
mightoccur: Francein the periodof the assignats; Europein the 1920s;
Greece, Hungary,and Chinain the aftermathof WorldWarII.
I
In the
1980s, the inflation net is cast wider. Israel had its high inflation
experience in the early 1980s. Eastern Europe, in its transition to a
marketeconomy, is currentlyhaving its share:Polandand Yugoslavia
have alreadyhad double-digitmonthlyinflation;Bulgariaand the Soviet
Union may be next. In Latin Americahigh inflationis the rule: Bolivia
in 1984-85;Argentina,Brazil,and Peruare now experiencingwild price
instability.In many of these cases, the fiscal consequences of a political
transition,a terms of trade shock, or a debt crisis can be offered as the
initiatingdisturbances.
Well over a thousand papers have been written about extreme
inflation, including early analyses by Costantino Bresciani-Turroni,
We are indebtedto our discussants, members
of the panel,
and to
Eliana Cardoso,
PhillipCagan,
Daniel Dantas, Hans Genberg,Danny Quah,
and Steve Webbfor helpful
commentsand suggestions. Miguel Cantillocontributedexcellent research assistance.
Financialsupportwas providedby grants from the National Science Foundationand
the AlfredP. Sloan Foundation.
- See Seligman(1921)for the experiencepriorto the FirstWorldWar;Cagan(1956),
Yeager(1981),Young(1925),Sargent(1982),DornbuschandFischer(1986),andHarberger
(1978,1988)for laterexperiences.
I
2 Brookings Papers on
Economic Activity,
2:
FrankGraham,andPhillipCagan.
Furthermore,interestgrows as more
is learnedabout the systematicpatternsof high inflation,especially
for
countrieswhereinflationis bothchronicandextreme, thoughlower than
the extravagantlevels studiedby Cagan.The formalizationof concepts
such as credibility and reputationhas also increased interest in high
inflationexperiencesby providing
cleaneranalyticalmodels.3The polit-
ical economy literature,too, has contributedto the debateby explaining
how hyperinflationsstart
up, why their terminationis so difficult,and
what ultimatelystops them.
This paperexaminesthese issues and addressestwo questions:
What
do we know about the process of high inflation-its sources, catalysts,
and explosive behavior?Whatuseful prescriptionsare there for stabili-
zation?
Extreme Inflation
Highinflationmeansdifferentthingsto differentpeople. In Germany,
it means 3 percent or more per year; in Mexico, 20 percent per year; in
Brazil, 15 percent per month; and in Argentina, 6 percent per week.
WhileCagandefinedhyperinflationas an inflationrate of 50 percentper
month, or 12,875percentper year, our thresholdis a moremodest 1,
percent per year.4In particular,we take extreme inflationto be rates
above 15 to 20 percentper month, sustainedfor several months.
In either 1988 or 1989, 35 countriesexperiencedannualinflationrates
exceeding 20 percent. Of these countries, 9 experiencedannualinflation
above 100 percent, and only 5 had annual inflationrates above 1,
percent.5Clearly,extreme inflationis not a commonoccurrence.A first
task, therefore,is to isolatethose factorsthatexplainwhy mostcountries
do not cross the thresholdto extreme inflation.
StabilizingFactors
A majorshock to the budget,the terms of trade, or the exchange
rate
is an essential ingredientfor high inflation.An event such as a political
- Bresciani-Turroni(1937);Graham(1930);Cagan(1956).
- See, for example,BarroandGordon(1983),Bruno(1990),andKiguelandLiviatan
(1990b).
Cagan(1956).
- See IMF, International Financial Statistics.
4 Brookings Papers on Economic Activity, 2:
Table 1. Annual Rates of Inflation in Selected High Inflation Countries,
1975-
Percent
Most recent
Country
1975-80 1980-85 1986-89 12 monthsa
Argentina 193 323 588 14,
Bolivia 17 611 15 15
Brazil 51 149 610 6,
Israel 61 195 19 20
Mexico 21 61 81 27
Nicaragua 19 54 3,130 10,
Peru 50 102 693 2,
Poland 7 32 90 994
Yugoslavia 18 47 343 1,
Source: IMF, International Financial Statistics, 1990 Yearbook and October 1990 editions, except 1989 data for
Nicaragua, which was provided by the Nicaraguan authorities.
a. The average inflation rate over the most recent twelve months for which
data are available for each country
is
shown.
financialmarkets. But this same indexationmay prevent the economy
from effectively absorbingshocks to relative prices and may leave the
economy proneto hyperinflation.Whenindexingadjustmentsare infre-
quent, there is a long lag between shocks and theirimpacton wages. By
contrast, when indexation occurs monthly or more frequently, the
automaticfeedbackof disturbancesto wages andprices, andbackto the
exchange rate, amounts to a rapid, vicious cycle of rising prices. The
financialadaptationalready in place-the easy availabilityof interest-
bearingassets-facilitates the flightfrom money.
The second possible scenario arises when a country with moderate
inflation suffers a rapid, uncontrolled explosion in inflation due to a
shock to the budget or the external balance. In table 1, Mexico could
provideanexample.
In this situation,anoutburst
of inflation
will surprise
money-holdersand may
find the authorities unprepared
for the rapid
adjustmentsthat the private
sector makes.
Finally, an extreme inflationcan occur in a country where inflation
has been repressed and where deficit finance has built up a monetary
"overhang."
In this case,
there is not only
the ordinary
flow problem
of
a budget
deficitfinanced by money
creationbut also a stock problem
in
that the money supply
is already
too high
relative to nominal income
at
controlled prices. Once prices
are freed up,
inflation gets rapidly
under-
way. In the absence of strongstabilizingfactors, it can become extreme
RudigerDornbusch,Federico Sturzenegger,and Holger Wolf 5
very quickly.The fear of this type of explosive inflationled to pervasive
monetaryreform in Germanyand elsewhere in the 1940s.6 Poland, in
1989, is a more recent example. The Soviet Union and Bulgariamay
providethe next examples.
A country's starting point is important because it bears on the
dynamicsof highinflation.Countrieswith chronicinflationtend to have
a financialstructurethatis adaptedto inflation;capitalflight,the resulting
exchange rate collapse, and the consequent explosive inflationmay be
stretched out over time. By contrast, in countries that have little
experiencewithhighinflation,primitivefinancialstructures,andentirely
unindexedfiscal institutions,an inflationaryshock can set the house on
firein no time. Becausethe financialstructureis so unadaptedto inflation,
therewill typicallybe an initialphase in which real balancesrise rapidly
as a result of deficit finance. If no interest-bearingassets are available
anddollar-holdingis uncommon,little adjustmentto realbalancestakes
place initially.Later, as escalatinginflationbecomes apparent,
the flight
frommoney into foreignassets acceleratesdramatically.
Are All Hyperinflations Alike?
High inflationexperiences have many common, systematicfeatures
includingdemonetization,growingdeficits,andacceleratingfrequencies
of wage
and price adjustment. High
inflations also differ from one
another.For example, the great inflationsof the 1920swere considered
extreme aberrations,which could be remediedby a returnto the gold
standard. By contrast, the current inflationin some Latin American
countriesremainsendemic, and has no certaincure.
The recent LatinAmericanexperiencediffersfromthose of the 1920s
in three ways. First, the inflationsin Latin America, and also in Israel
priorto the 1985 stabilization,have lasted longer. They have extended
overa decadeormorethroughthe intermittentbutpersistentacceleration
of inflation.Second, currentinflationrates remainfar below those of,
for example, Germany in the second half of 1923, where inflation
averaged4,551 percentper month.Third,in recent times there has been
a stop-gopatternof inflation-temporary stabilizationsare followed by
majorblowups. The temporarystabilizationsinitially seem to prevent
- DornbuschandWolf(1990).
Rudiger Dornbusch, Federico Sturzenegger, and Holger Wolf 7
Figure 1. Adjustment to a Sustained Increase in Money Growth
Inflation
Actual
" rate of
inflation (IT)
~- __ Rate of
moneygrowth(,u)
, Expected
,' rate of
Time
threeissues: the expectationsmechanism,the assumptionsaboutmoney
growth, and the assumptionsabout the determinantsof inflation.Criti-
cism of the expectations mechanism,notablythe rationalexpectations
attack, has focused on the serial correlationof forecast errors in the
adjustmentprocess. An alternativemodelwouldemphasizefullyrational
expectations.9Whenperfectforesightis assumed, two differenceswith
the previous analysis emerge. First, the price level is allowed to jump.
Second, actualand anticipatedinflationare equal,
r = r*.
In this forward-lookingmodel of inflation,an increase
in the growth
rate of nominalmoney immediately
raises both the price
level and the
rate of inflation. Thus, real
balances instantly decline
to the level
appropriateto the highergrowthrate
of nominal money
and prices. Full
adjustmentto an increase
in money growth
occurs immediately;
there
areno protracteddynamics.Anticipated,future
money growth
stillleads
- Sidrauski(1967),SargentandWallace(1973a),Black(1974),Sargent(1977),Salemi
andSargent(1979),andFlood andGarber (1980).
8 Brookings Papers on Economic Activity, 2:
to ajumpin the currentpricelevel, but now inflationrises graduallyuntil
money growthincreases.
A New Approach
We departfromthe Caganmodel and the relatedliteraturein several
ways. Although the rationalexpectations formulationis theoretically
attractive,it does not reflectthe facts. Increasedmoney growthinitially
tends to raise real balances; only later does inflation build up and
overshoot for a while.
Figure 2 shows the exampleof Peruafter 1985.
In our model, we emphasizethat prices are set ratherthan determined
by an invisible, money market-clearinghand. The price setting, in turn,
is dependenton the prevailinginflationaryregime.
A second difference in our model concerns financialmarkets. The
assumptionthat money is
the only asset
is patently misleading.Some
countries do have domestic financialmarkets. When they
do not,
the
dollar usually plays a conspicuous role as an alternative asset. The
availabilityof alternativeassets influencesthe dynamicsof inflation.
The third difference in our model concerns the process of money
growth,and the motivationa governmentmay have to create or accom-
modateinflation.As a theoryof inflationaryfinance,the Cagan-Chicago
traditionhas emphasized that governmentsuse inflationas a rational
means of raising revenue for deficit finance. This approachhas led to
models of optimalinflation,which balancerevenue considerationsand
various constraints or costs associated with inflationaryfinance.1'In
their simplestform, such models predictthat a governmentmaximizes
seigniorageby choosing the rate of inflationat which the elasticity
of
real money demandwith respect to anticipatedinflationequals unity,
or
the peak of the Laffercurve.
MartinBailey has questionedthe use of the seigniorageapproachas
a positivetheoryof extremeinflation.
12 Because taxationis analternative
to money creationandbecause the marginalcost of taxationis moderate
relative to the welfare costs of extreme inflation,high money growth
(and therefore high inflation)strategies are not optimal. In our view,
- FriedmanandSchwartz(1982)set out this pattern.
- See Cagan(1956),Bailey(1956),Friedman(1971),
Barro (1972a,1983),
andBruno
(1990).
- Bailey(1956).
10 Brookings Papers on Economic Activity, 2:
economy approach that analyzes the difficulty in bringing about a
balancedbudgetand illustrateshow extreme inflationis a negative sum
game. With the seigniorageapproachone would expect stable, steady-
state inflation.The politicaleconomy approach, however, may explain
why inflationexplodes, only to be stabilized later, and why rational
governmentsmake U-turnsin economic policy. 13
In conclusion,the view thatinflationis abouttoo muchmoneychasing
too few goods is not totally wrong, but
it fails to give a useful idea of
whatin fact happens.First, the influenceof inflationon the budgetdeficit
is a critical element in the inflation. Second, the inflationprocess is
significantlyaffected by the adaptationof financial institutions. The
emergenceof new financialproductssuch as overnight,interest-bearing
depositsaffects
the government'sability
to collect seigniorage
andhence
shapes
the inflationprocess. Third,
the inflationprocess
is drivenin part
by the pricingrules of economic agentsin the goods andfactor markets.
A shorteningof adjustmentintervalsbecomes a factorin inflationthat is
independentof aggregatedemand. Finally, as we discuss below, the
external sector can play a critical role in the inflationprocess because
real exchange rates are not constant. Indeed, the assumptionof pur-
chasingpower parityis empiricallyunwarranted.As a result, shocks to
the real exchange rate can precipitatehighinflation.
Inflation, Money Growth, and the Budget
In this section we discuss the linksbetween the growthrateof nominal
money, the budget deficit, and the rate of inflation.We emphasize two
factors. First, money growth is endogenous;money passively finances
budgetdeficits. Second, the realbudgetdeficitis not exogenous because
inflationaffects the real level of revenues.
The inflationaryequilibriumis described by monetary equilibrium
and the link between money creationand the budgetdeficit: 14
(4) MIP = OY,
- For explorationsin this direction, see Alesina and Drazen (1990), Cukierman,
Edwards,and Tabellini(1989), Orphanides(1990);and, in anothercontext, Fernandez
andRodrik(1990).
- We note thatthe budgetdeficitincludesthe "quasi-fiscal"deficitthatarisesfrom
the credit and foreignexchange operationsof the centralbank. Also, money issued by
state and local governments,by corporations,andeven by privateentrepreneursis often
significant.The seigniorage shared by the banking system and other money issues
necessarilyincreasesthe inflationrate associated
with any given budgetdeficit.
Rudiger Dornbusch, Federico Sturzenegger, and Holger Wolf 11
where 0 is the budgetdeficitas a shareof realGDP, Y.Supposenow that
velocity is a linearfunctionof inflation,V = cx+ 131r.Imposingmonetary
equilibriumand the steady-stateconditionthat velocity be constant (so
that inflationequals nominalmoney growth), we obtain an expression
for steady-stateinflation:
(5) rr= cX0/(1 - (0).
It is apparentfrom the functionalform that inflationis an increasing,
nonlinearfunctionof the budgetdeficit. 15
The interactionbetween the budget and prices-both their level and
theirrate of change-may go in one of two directions. Eitherfiscal drag
and the reduction of
the real value of debts dominate,
in which case
inflation reduces deficits. Or, inflation worsens deficits because of
collection lags that reduce real revenues. During
the hyperinflation
of
the 1920s,it was generallyacknowledgedthat inflationreducedthe real
revenue of governments and that, by implication, deficits would be
smallerif inflationcould be reduced. 16 The adverse response of deficits
to inflationis now known as the Tanzi effect. 17
We identifyfive channelsthroughwhich inflationaffects the deficit.
-Lags in collection in a less thanfully indexedtax system lead to the
erosion of real revenues, not only automatically, but also from the
additionalerosion thatresults frommanipulateddelays in payment.
-Tax compliance deteriorates, and so the tax yield of a given tax
structuredeclines. This effect is a by-productof inflationand public
demoralization;it is not a mechanicaleffect.
-The real value of payments on long-term debt, fixed in nominal
terms, declines. This mechanism was importantin the 1920s but is
marginalin the experiences of the 1980s.
-With increased inflation and unchanged indexation intervals for
wages, the average real public sector wage between adjustmentdates
declines.
-Nominal interest payments on short-termdomestic debt come to
- An alternativespecificationis the Caganform which yields 0 = w-reX,.
Because
the right-handside has the Laffercurveproperty,thereare two equilibriaunlessthe deficit
is too largeto be financedby moneycreationat constantinflation.See Evansand Yarrow
(1981),Cardoso(forthcoming),Kiguel(1989),andespeciallyBrunoandFischer (1990)on
the implicationsof the Laffercurve.
- See Bresciani-Turroni(1937),Graham(1930),andLeagueof Nations (1946).
- See Olivera(1967),Dutton(1971),andTanzi(1977, 1978).
Rudiger Dornbusch, Federico Sturzenegger, and Holger Wolf 13
Table2. Peru:PublicFinanceData, Expressed
as Percentof GDP,
1985-
Percent
1985 1986 1987 1988 1989
Tax collection 13 11 9 7 4
Publicenterpriserevenue 26 18 14 12 10
Budgetdeficit 6 10 12 14 11
Source: Unpublished World Bank data.
tax revenue collected today decays exponentially
with a mean lag
of a.
In that case, the deficitratio
is
(6) 0
g - v/(1 + C T),
where g denotes real governmentspendingas a fractionof GDP, and a
can assume a value between zero and infinity(the closer to zero, the
more currentthe collection). Equation 6 shows that the more current
the tax collection, the higheris the real revenue. It also shows that the
higherthe rateof inflation,the lowerthe realyieldof a giventax structure.
For a given collection lag, an extreme increase in the rate of inflation
will basicallywipe out revenue.
Given the velocity equationand the noninflationarydeficit (g
v),
the longerthe averagelag in tax collection, the higheris the steady-state
rateof inflation.Moreover,giventhe nonlinearities,
even small increases
in the deficit can producelargechanges
in inflation.
Equation 6 focuses on the case of a constant-inflation history. Yet,
when inflationhas been rising, the currentreal value of tax collection
will depend on the history of inflationacceleration. The sharper
the
accelerationbecomes, relative to the lag
in tax collection,
the further
realrevenuefalls comparedto the constant-inflation-rateformula.In the
simulations below,
this complicationemerges as
a critical destabilizing
element.
To summarize,the endogeneityof the money stock is an important
featureof hyperinflation.In economieswheremoneycreationis the only
means of deficit finance, the budget deficit becomes a principaldeter-
minant of money growth.
In addition,
faster money growth
leads to
- It is apparentfromthe nonlinearitiesin equations 2 and 3 thatequilibrium
need not
be unique.
14 BrookingsPapers on Economic Activity, 2:
higherinflationwhich, in turn,determinesthe deficit.The interactionis
highly nonlinear,which helps explain the sharp escalation of inflation
that can occur once a certainthresholdis crossed. A simulationmodel
presented below displays that characteristicin a setting where pricing
decisionsandthe velocity responseto inflationareoptimized.It becomes
evident that extremeinflationemergesin those cases where the govern-
ment has lost controlover money growthbecause the deficitis financed
automatically.To make things worse, the deficit itself responds posi-
tively to the rate of inflation.
Pricing
The interactionbetween inflationand the frequency
of price adjust-
ment represents the second building block of inflationarydynamics.
Economic agents incur costs in adjustingprices. Thus, the optimal
frequencyfor adjustingprices depends on these "menu costs" and the
trendrateof inflation.Thefrequencyof adjustmentbecomes endogenous
in a high inflation economy, with the shortening of the adjustment
intervalacceleratinginflationwhich, in turn,leadsto a furthershortening
of adjustmentintervals.Withoutmonetaryaccommodationthis process
could not go far. But, as already noted, money growth is endogenous
through budget deficit finance. Thus, the shortening of adjustment
intervals becomes another endogenous driving force in the inflation
process.
Strategic interactionamong price setters is an importantfeature of
the growing literatureon optimal price adjustmentin stochastic infla-
tionaryeconomies.20We are interestedin the fact that the frequencyof
adjustmentaccelerateswith inflation.We also wantto learnmoreabout
the thresholdsof inflationat which the shorteningof pricingintervals
occurs.
The Firm and the Industry
Individualfirms face the microeconomic decision of how often to
change prices. Considera loss functionfor an individualfirmin which
- See Barro (1972b), Mussa (1981a), Sheshinski and Weiss (1983), Rotemberg (1983),
Cecchetti (1986), Benabou (1988), and Diamond (1990) for models of optimal pricing rules.
16 BrookingsPapers on Economic Activity,
2:
Relative Prices
Whenaggregateinflationis moderate,firmandindustryprices change
infrequently,perhapsonce a year.
Withhigherinflation,priceadjustment
may be semiannual, quarterly, or ultimately monthly, weekly, even
daily. In the limit, indexationto the dollarbecomes the only way a firm
can judge its relative price position given the prohibitive costs of
ascertainingthe currentprice level. Because the pricing by firms and
industriestranslatesinto an aggregaterate of inflation,theirprices have
spillovereffects: on the pricingof labor,includingofficial
minimumwage
and public sector wage setting; on the adjustmentof prices for public
services; and, importantly,
on the exchangerate.
Public sector price adjustmentshouldfollow the same logic as firms'
pricingdecisions, though it rarely does. The data in table 2 give a fair
idea of the fiscal effects of misaligned relative prices in an extreme
inflationcountry, Peru. As inflationaccelerates, both tax revenue and
publicenterpriserevenue decline as a shareof GNP.
It is less evident that menu costs are importantfor exchange rate
adjustments.Therefore,governmentsmightchoose to move exchange
rates almost continuouslyin orderto maintaintheirreal value. Indeed,
several countries, for example Brazil in the 1970s, have done just that.
But, as with public sector prices, exchange rates are a seemingly
attractivemeans of dampeninginflation
in the shortrun, and so they are
abusedfor
that end.
Evidence on Pricing
The generalpattern
of price
adjustmentis shownbelow. We note that
with increasinginflationall price setting speeds up and slower-moving
prices would catch up.
Speed of adjustment
Frequent
Intermediate Slow
Exchangerate Publicprices Wages
WPI CPI
The increasingsynchronizationthat emerges during
a shift to high
inflationcan be seen from the cross correlations
of exchange
rates and
prices. In countries
withhistoriesof persistentinflation,
the linkbetween
Rudiger Dornbusch, Federico Sturzenegger, and Holger Wolf 17
Table 3. Correlation between Prices and the Current Exchange Rate,
Selected Inflationary Episodes
Consumerprices
Average
Ahead two Lagged inflation
Country
and time period
months Current two months rate (percent)
Bolivia
1980:1-1983:1 0.233 0.531 0.472 5.41a
1983:2-1986:2 0.870 0.
Israel
1980:1-1982:9 0.611 0.
6.20a
1982:10-1985:7 0.
0.999 0.839 11.
Argentina
1985:7-1989:1 0.878 0.993 0.866 10.
1989:1-1989:7 0.235 0.968 0.003 49.
Brazil
1960:1-1969:12 0.945 0.988 0.
1980:1-1985:2 0.
0.993 0.787 3.
1985:3-1990:
0.709 0.987 0.626 16.88a
Source: Authors'own calculationsand InternationalFinancialStatistics. The data are monthly,and prices are
the consumerpriceindexesfor each country.
a. The inflationrateis the averageover the periodsindicated,except wheredatawere available:the firstinflation
figurefor Bolivia is averagefrom May 1980 to January1983;the first figurefor Israel is from October 1980 to
September1982;and the last figurefor Brazilis fromMarch 1985 to April 1990.
the dollarand domestic prices is well developed. An increase in the rate
of inflationresults in a shorteningof the adjustmentlags, mirroredby a
tendency for the correlationsto become centered on the contempora-
neous value. By contrast,in countrieswithouta historyof exchangerate
indexation,acceleratinginflationleads to the developmentof indexation
mechanisms,resultingin a uniformincreasein correlationsacross leads
andlags.
In table 3,
we divide inflationaryepisodes
into several parts and
present the correlationsbetween the currentexchange rate and both
currentprices and prices with leads and lags. The data broadlysupport
our hypothesis.
For Braziland Argentina,
it is interesting
to note that as
inflationincreases over time the correlationwith leads and lags falls off
sharply.
Inflation
and Price Adjustment
An importantaspectof pricinginvolves the basison whichadjustments
are made. At moderaterates of inflation,adjustmentwill be based on
past inflation.As inflationrises, the adjustmentintervalshrinks,though
Rudiger Dornbusch, Federico Sturzenegger, and Holger Wolf 19
usually based on formal or informal backward-looking indexation. This
practice implies that when inflation accelerates, and the indexation
interval does not decline nor are base wages raised, the average real
wage declines. Reuben Kessel and Armen Alchian refer to this phenom-
enon as the wage-lag hypothesis and contest its existence.24 At issue is
whether the real wage does decline in periods of accelerating inflation.
And, if so, should the decline be attributed to a fall in the equilibrium
real wage of a competitive labor market, or to features of wage setting
that show insufficient flexibility in the face of accelerating, and perhaps
incompletely anticipated, inflation?
The real minimum wage in Brazil, as shown in figure 3, provides an
example.25 The nominal minimum wage is officially set on the basis of
accumulated price increases, and applies to a significant part of the
Brazilian labor market, including household workers and the informal
sector. At the end of the 1970s, with annual inflation rates of 40-
percent, nominal wages were adjusted only once a year. In 1979, the
adjustment became semiannual (in May and November) even though
the rate of inflation had risen to more than 200 percent. By 1986, the
frequency of minimum wage adjustments became discretionary depend-
ing on which stabilization program, which invariably included wage-
price controls, was in effect. From 1989 to the present, adjustments have
been monthly or, at most, bimonthly. They continue to be backward-
looking, in that the adjustments equal the inflation that has accumulated
since the last readjustment.
In every high inflation, there is pressure to adjust wages more
frequently and more fully. The following quote from the German expe-
rience of 1919-24 characterizes the issues:
Therewas no lack of attemptsto make the index numberthe sole determining
factor in wage negotiations. The battle for the sliding wage scale began in
Germany in the middle of 1919, and never really quite stopped until the
introductionof the system of gold wages. In the continuedstruggleto maintain
the level of realwages, the workersshoweda comprehensiblepreferencefor the
system of complete automatic adaptation of wages.... The employers almost
unanimouslyrefused the sliding scale, which they feared would lead to rapid
andcompletedisaster,both monetaryand economic....
- Kessel andAlchian(1960).
- The same patterncan be shown for wages in manufacturing.Accordinglythe
propertiesarenot peculiarto administeredwages. Pazos(1972)commentedon this seesaw
patternof realwages.
- Stizler(1924,p. 648).
20 Brookings Papers on Economic Activity,
2:
Figure 3. Brazil: Real Minimum Wage, 1977-
Real minimum wage (index,
January 1978 - 100)
130
120-
110 1
100
90
80-
70-
60-
50-
,. , , , , , , ,
1978 1980
1982 1984 1986 1988 1990
Year
Source: BancoCentraldo
BrasilBoletim Mensual and Conjuntura
Economica, variousissues.
In the extreme stages of the hyperinflation,indexation
was complete.
The only issue was whetherthe index
numberof Monday'sprices,
which
was publishedon Wednesday, could
still be useful for wage payments
madelater in the week.
Why does the labor market accept
so much variability
in the real
wage? Since much of
the variability
is anticipated,
and since more
frequent adjustmentsof money wages (as
opposed to more frequent
payments) do not
have a resource cost beyond
checking the
latest
inflation figure,
there is a puzzle.
With a given
frequency of wage
adjustments,higher
inflation automatically
depressesaveragereal wages.
As such, firmsmightbe tempted
to resist a shortening
of the indexation
interval. But why should
workers not insist on higher
base wages,
and
what stops firmsfromcompeting
with moreeffective wage plans?
While
- See Stizler(1924,p. 655).
