Public Disclosure Authorized

Policy,Research,andExternalAffairs

WORKING-PAPERS
MacroeconomicAdjustment
and Growth

Country Economics Department

The World Bank
August 1991
WPS 752
Public Disclosure Authorized

Inflation and Stabilization

in Yugoslavia
Public Disclosure Authorized

Robertode RezendeRocha
Public Disclosure Authorized

A successful stabilization program in Yugoslavia requires more

political resolve about wage indiscipline and loss-making enter-
prises than was observed in 1990. But the ultimate question is
whether stabilization can succeed without a comprehenisive
privatization program.

ThePolicy. Research, and ExtemalAffairs Complexdistnhutesl'RI' Working ladprsto disseniiniat the findings of work in progrcss and
to cnoouragethe cxchangeof idcas among lanik stafl' and all others... erested in development issucs. These papers carry the names of
the authors, rflect only theie views, and should be used and cited accordingly.'Ithefindduigs,interpretations, and conclusions arc the
authors own. They should not be atibuted to the World liank, its Board of l)irectors, its managmncnL, or any of its member countries.
 Policy,Research, andExternalAffairs

MacroeconomicAdjustment
and Growth

WPS 752

This paper -a product of the Macroeconomic Adjustment and Growth Division, Country Economics
Department- is part of a larger effort in PRE to examine the problems of transition faced by reforming
socialist countries and to contribute to the Bank's policy dialogue with these countries. Copies are available
free from the World Bank, 1818H Street NW, Washington DC 20433. Please contact Lanha Ly, room N9-
083, extension 37352 (35 pages).
Rocha exarnines the main reasons inflation was initially successful in fighting inflation. But
accelerated in Yugoslavia in the 1980s and it became clear in the course of the program that
reviews past and current attempts at stabilization. other losses had not been removed. Pressures to
finance enterprises and avoid a liquidity crisis in
He shows that inflation in Yugoslavia shares the financial system resulted in a relaxation of
common elements with inflation in other highly monetary policy in mid- 1990 and a revival of
indebted countries, despite appearances other- inilationary pressures. Attempts to reimpose
wise. These common elements include a large monetary control met considerable difficulty at
transfer of resources abroad unmatched by an the end of tlheyear, includinlg a bizarre episode
internai adjustment, resulting in a large internal of expansion of central bank credits without the
redistribution of real resources through inflation. board of governors' approval.

Yugoslavia differs from other countries in It also became clear that the fiscal compo-
that these internal conditions are not transparent. nent was not consistent with other elements of
Instead of an open fiscal deficit, there were the program. It was clearly not enough to
complex interactions among enterprises, com- finance a social program of the magnitude
mercial banks, and the central bank, involving, required had loss-making enterprises really been
among otner things, the absorption and servicing lorced into bankruptcy and also to cover the
of a large stock of foreign exchange liabilities by needs of the bank restructuring program. Seen
the central bank. from this angle, the Yugoslav program of 1990
resembles other heterodox programs that had
Other factors contributed to the sharp initial success in reducing inflation but later
acceleration of inflation at the end of the eighties faltered because of the insufficiency of the fiscal
- especially a large real devaluation in mid- adjustment.
1988, when an indexed economy drove inflation
to a much higher level. In 1989, a preemptive At the same time, the events in the second
explosion of real wages added fuel to inflation's half of' 1990 also indicate that, for a stabilization
fire. program to succeed in Yugoslavia, there must be
much greater political resolve to cope with wage
Rocha argues that the failure to correct indiscipline and loss-making enterprises than
hidden losses in the economy was the main was observed in 1990. And the question
reason various stabilization attempts failed in the remains whether financial discipline can be
1980s. The 1990 program was the first to imposed in the system only at the macroeco-
recognize the existence of those hidden losses nomic level and without introducing private
and the need for fiscal correction - although it ownership of capital. The ultimate question may
also introduced other elements to cope with be whetlherstabilization can succeed without a
inflationary inertia. The program succeeded in comprehlensiveprivatization program.
eliminating the central bank's own deficit and

The PRE Working Paper Serics disseminates thc findings of work under way in the Bank's Policy, Research, and Extemal
I Affairs Complex. An objective ofthc scrics is toget thesc fin(ings ot qu(luickly,
even itprcsentalions arc Icssthan fully polished.
The findings, interpretations, and conclusions in thesc papers do not necessarily rcprcsent official Bank policy.

Produced by the PRE Dissemination Center

 Inflation and Stabilization in Yugoslavia

by
Robertode Rezende Rocha*

Table of Contents

1. Introduction 2
2. Enterprise Losses, Central Bank Deficits, and Inflation 5
3. Exchange Rate, Wages, and Prices 17
4. Past StabilizationAttempts and the 1990Program 25
5. Conclusions so
References 33

* I am grateful to Barry Bosworth, Simon Commander, Vittorio Corbo, Wei Ding,

Rudiger Dornbusch, Manuel Hinds, Mitja Gaspari, Miguel Kiguel, Neven Mates, Branko
Milanovic, Jeffrey Sachs, Boris Skapin, William Tyler, Milan Vodopivec, and Steve
Webb for useful comments on earlier drafts. Conversations with Valimir Bole were also
very helpful.
 2
1. Introduction.

The reversalof externalfinancingflows during the 19808forced Yugoslavia

to undertake drastic measures to balance its external accounts. The external
adjustment consisted primarily of large real exchange rate devaluations and,
initially,quantitativerestrictionson imports. Although these measures were
able to generate increasing current account surpluses, economic activity
stagnatedand inflation acceleratedalmost continuously(figure 1 and table 1).
Moreover, the absence of fiscal imbalances in the economy (figure 1) might
suggest that the nature of inflation in Yugoslavia is entirely non fiscal.

Table 1
Annual Rates of Inflation (CPI) in Yugoslavia, 1980-90
(in S)
80 81 82 83 84 85 86 87 88 89 90
Dec. 37 36 33 60 53 75 92 169 245 2685 120
Aver. 31 39 31 40 54 72 90 120 194 1240 588
Sources IFS

While it is true that many hyperinflationepisodes are triggeredby balance

of payments difficulties and large exchange rate devaluations, the complete
absence of fiscal imbalancesfrom the overall picture in the case of Yugoslavia
is intriguing. Indeed, even the "balance of payments view of inflation"
recognizes the role of fiscal deficits in the determination of inflation,
although such a role is assumed to be less central than under the "fiscal view
of inflation". In the former, the exchangerate assumes the central role, while
fiscal deficits contribute to inflationmostly through endogenous interactions
with the exchange rate and the inflationrate itself (Dornbusch1987; Dornbusch
and Fischer 1986; Liviatan and Piterman 1986; and Montiel 1989).
 Figure 1
VariablesFor Yugoslavia,1977-89
SelectedMacroeconomic

Exporto and Imports,1977-89 RealExohangeRate,1977-1989

(BiIllonsof US$) (1980a 100,
170

14 . . ........ , ................................. .e..... . ........ ......-....tc

It4 _.... "I. ........ .......... ........... ........ 1'- ......... .~...... .................... ........... ''.

tg, .. . ............. .. ............. .......... '''-t40 .. ,_,,_,_,-I_._..... ............ _.._... .. ..................

IC *-*tOC _ ,,

..... .......... .... ............... ......

f~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~o
......
r ne 19at
SrO IM =I8 '"M nt4 w08 tf Ss MeB INSf ..
r- . {10 .___ _ _ _ ______bs...........

- EOlDrtto +- Im9orto K
S
_~~~~~~~~~~~~~~~~~~~~~~~~~~~~0
1l77 tore 1970
__. . . .........
98
.... ._... ,-......
198
.
198l
.
2 198
.
1984
.
t08980I1987
. . .
198
.
¶98

Current Account,1977-89 Inflation R"te

(Billionsof USO) (OPI,'warly Chan" In %l

10000w Loasss

..0I.....
.........I.........
_.
4 _ _~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.__..._ 0 ._.......
__ ._.... . .......

2_,_,.__
__~~.. .. ... .. .. .. .... ..
=...................................

C -2 ....... ........................
10sS.
. _ _ _

...... ........ .
4 . ............. . . . ...
. . ....
........

Pubilc Sector Surplus.1977-89 |Real GSPGrowth,1977-89

( of asp) (In % pa.)
4~~~~~~~~~~~~~~~~~~

1977 19781979 90 1881e

1882 198 188 =198 Me8 1987 198 18917 9817 90 9118 9818 98 9818 9818

o-A~~~~~~ t - ...........................
4__\_

.2
i1 Z 8tl t9Bo= W
= 102 tD
Ioa W4 t0 %89100t t9B1
198 39S le"7 reo10tt t93 toai im8 %as h@ tees we mWtm 19S9
 4

The paper showsthat the accelerationof inflationin Yugoslaviaafter 1982

resembles in part the phenomenonof inflationin other high indebted countries,
despite initial appearances. The resemblance lies on three common elements.
The first is the adverse impact that large real devaluationshad on the major
holders of foreign liabilities--inthe case of Yugoslavia, the enterprisesand
the central bant:. The second element was the lack of internal adjustment by
enterprisesor the public sector to match the large transferof resourcesabroad.
Instead, enterprise losses increased through the 1980s, partly as a result of
the real devaluationsand increasingproblemsof overstaffing. The third element
was the internal generation of resources through inflation.
Inflation in Yugoslavia seems to differ from this rather familiar story
due to the lack.of transparency of internal conditions. Instead of an open
fiscal deficit, there were complex interactionsbetween enterprises,commercial
banks and the central bank, involving,among other things, the assamptionof a
share of debt service payments by the latter. The situation was further
complicatedby the existence of a large stock of foreign exchange depositsheld
by residents in t0- domestic financial system. The central bank also absorbed
these deposits in order to protect the enterprises and commercial banks from
greater financialdifficulties,thus increasingfurther the level of its quasi-
fiscal expenditures.
While the financing of hidden losses was a permanent source of monetary
expansion and inflation through the 1980s, the paper also assesses the
contributionof non-fiscal factorsto the sharp accelerationof inflationat the
very end of the decade--therate of inflationincreasedmore than tenfoldbetween
mid-1988 and end-1989. More specifically,the paper assesses the impact of a
large real devaluationin mid-1988 in the contextof an indexed economy,and also
the impact of an explosion of real wages in the period preceding the
stabilizationprogram of 1990. The paper presents some econometric evidence
indicating that these factors became indeed more important at the end of the
 5
decade. 1

Finally, the paper reviews the performance of stabilizationprograms in

Yugoslavia,and assessesthe perspectivesof the stabilizationprogram of January
1990,the first programthat recognizesthe existenceof hidden lossesin the
economyand the need for a fiscalcorrection.
The paper is organizedas follows. The second section examinesthe
contributionof "fundamentals"to the accelerationof inflation in the 1980s.
centralbank
financialimbalances,
It examinesthe linksbetweenenterprises'
deficits, and inflationaryfinance. The third section analyzes the relationship

betweenexchangerates,wages and pricesduring the 1980s,payingparticular

very
attentionto the two last yearsof the decade,when inflationaccelerated
rapidly.The fourthsectionexaminespastand presentattemptsat stabilization,
with focus on the 1990 program. Finaily, the fifth section draws some
conclusions.

Losses. Central Bank Deficits and Inflation.

2. Enterprise
Otherstudiesprovidea detailedexaminationof the causesA e.fterprises'
insideYugoslavia(Knight1984;Konovalov1989;
losses,and their distribution
and Saldanha 1989). For the purposesof this paper highlightingthe main
findingsof thesestudiesis sufficient.Excessivepersonneland the setting
of wages by criteria completelydivorced from productivitystand out as one of
in
the major causesbehindpoor financialperformance.OtherX-inefficiencies
2. The lossesalso reflect
the use of materialinputsare also a major cause
investmentdecisionsmade at very distortedrelativeprices and without any
serious considerationabout potentialdemand. These investmentswere undertaken
mostly during the 1960s and 19709, and financedin good part by foreign credits.

I See Rocha (1989),for an analysisof the role playedby the finaLacial

systemin the Yugoslavinflation,Boleand Gaspari(1990)for a detailedanalysis
of policyregimesduringthe 1980s,and the earlierstudiesof Mates (1987),
Hencinger(1987)and Gaspari(1988).
per unit
2 For instance, Konovalov(1989)reportsthatenergyrequirements
of outputwere twiceas high as in westernenterprises.
 6
Finally, the distorted tax system which prevailed in Yugoslaviawas also aS a
cause of enterprise losses. For one, excessive payroll taxes increased
significantly the cost of labor to the enterprises. Also, the base of the
enterprise Income tax was defined as the income before the payment of wages, and
enterprises paid income taxes even if they generated net losses.
As shown in table 2, enterpriselosses increased from around 2.0 percent
of Gross Social Product (GSP) in the early 1980s, to approximately6 percent of
GSP in 1987-88 and 15 percent of GSP in 1989. These numbers are not indicative
of cash losses, since they includenon-cash expendituressuch as depreciation,
and revaluationof stock items. In addition,they are also affectedby frequent
changes in accounting rules, including the adoption of inflation-adjusted
accountingin 1987. However,they indicatea deterioratingtrend that 's highly
probable.

Table 2
EnterpriseLosses as Shares of GSP, 1981-88
1981-84 1985 1986 1987 1988 1989
2.1 2.8 3.0 6.6 5.7 15.0
Sources Social Accounting Office.

The real exchange rate devaluationsof the 1980.were one likely cause of
the increase in enterprise losses shown in table 2. These real devaluations
increasednot only the real domesticcost of imported inputs, but also the real
flows of interest payments on Yugoslavia'sexternal debt, 80 percent of which
was initially held by enterprises (the remaining 20 percent with the central
bank).3 Another major cause of increased enterprise losses is likely to have
been the increasingproblem of overstaffingobservedduring this period. While
GSP growth rates declined from 7 percent in the 19709 to 0.7 percent in the
1980s, the rates of unemploymentincreasedvery slightlybetweenthe two decades-

3 The full impact of real devaluationson interestpAvments was, however,

reduced by the transfer of some debt service payments to the central bank.
 7
-from 12 to less than 14 percent. As a result, the amount of excess labor in
Yugoslav enterprises is estimated to have increased to 20 percent of the labor
force at the and of the 1980s (Mencinger 1989). The lack of adjustment of
Yugoslav enterprises is further illustratedin figure 2, which shows the absence
of a relationshipbetween changes in the rate of unexLloymentand output growth
rates (the Okun's law) in Yugoslavia, in contrastwith the US and German cases
(see Gordon 1984; Okun 1970)4.
Finally, changes in the tax structureduring the 1980s had a very adverse
impact on enterprises, since they amounted to a shift from indirect taxes to
heavier payroll and enterprise income taxes--the share of these two taxes in
total tax revenues increased from 50 to 67 percent between 1983 and 19&9 (Ding
1990 and Mates 1991). Thus, these changes increased the costs of excess lab"r
in Yugoslav enterprises.
The existence of enterpriselosses raises the questionof how these losses
were financed. During the 1980s there were two basic sources of financing. The
first comprisedvarious forms of inter-enterprisefinancing,includingvoluntary
transfers from profit makers to loss makers inside the same industraalholding,
and forced transfers under various forms--para-fiscaltransfers and inter-
enterprisecredits and arrears. The second source of financing complrisedbank
credits at subsidizedterms. Of course, the fact that the enterprise-controlled
banks maintainedthe policy of credit subsidieswas only sustainedby the payment
of even more negative real interest rates on domestic deposits (figure 3)5.
Note that only at the end of the decade real interestrates became positive, due
to the indexationof time deposits of three months and abovs. The consequences
of this measure are examined in section 4.

4 Figure 3 shows simple OLS regressionlines for the 1975-88period for the
three countries.
5 Figure 3 show the monthly averagesof realizedreal rates on one year time
deposits,computed by rt = (1 + it)/(l + at+,), where r and it are the real and
nominal interest rates at t and lrt+lthe inflation rate between t and t+1.
 Figure 2

Okun'e Law in SelectedCountries, 1975-88

UNITED STATES
7.5.

5.0 .

Ch5.0.0
anges in thxe U ne mploy ment Rate
1.5

-5.0

-L -2 1 -0 3

7O.0

8.0-
Ch anges in the Unemploynment Rate

.0-i
YUGOSLAVIA
10.0 .

-2.5
-n.s -o,0 035 1.o 1.5
Changes in the Unemployment Rate
 9

Figure 3

Average Real Rates on 1 Year De osTts

CkfltntlY AVOrt4Go. In fs p.a.3

-10

-15

-20 _S_~~~~~~~~ I

-25

- 30

-35 - , , , ,._
1990 1991 1992 1093 194 4985 1995 1987 1999 1999

While the provision of bank credits at subsidizedterms was instrumental

in financing enterprises'imbalances,the stock of bank credits would not have
grown at high rates without a commensuraterate of monetary expansion, most
particularlyof base money. Therefore, an examination of the primary sources
of base money creation is in order.
The rapid expansion of base money during the 19809 diu not result from
direct financingof enterpriselosses by the central bank. However, it resulted
in great part from the central bank's attempts to protect enterprises and
commercialbanks from even greater financial problems. First, after 1982 the
central bank increased the provision of selective rediscounts (exports,
agriculture,equipment,etc.) at very subsidizedterms. Second, the centralbank
absorbed and serviced a large volume of foreign liabilitiespreviouslyheld by
the enterprises and O.1 banks.
More specifically,after 1982 severalenterprisesexperienceddifficulties
in servicing their foreign liabilities. As a result, the central bank assumed
a share of enterprises' foreign liabilities,mostly of those located in less
 10

developed regions. Second, the centralbank aloo absorbedand servicedthe whole

stock of foreign exchange (FX) deposits in the commercial banke. These IX
depositsare held mostly by residents,are not part of Yugoslavia'sexternaldebt
(table 3), and accounted for two thirds of the central bank's foreign
liabilities.

Table 3
External Debt and FX Deposits, 1980-89
(In % of GSP)
1981 1983 1985 1986 1987 1988 1989
External Debt 30.0 32.1 44.1 33.0 30.7 32.1 25.0
FX Deposits 12.5 13.3 19.1 17.5 17.5 17.8 17.4
Sources National Bank of Yugoslavia.

The absorptionof the stock was done through the introductionof a foreign
exchange insurance scheme that involved the transfer of FX deposits to the
central bank in exchangefor an equivalentamount of low interestDinar credits
from the central bank. The purpose of the schemewas to shield commercialbanks
against foreign exchange losses, while also allowing them to keep granting
subsidized credits to enterprises.
The fact that the central bank started servicinga large volume of foreigr.
liabili.zies
while charging negative real interest rates on all its credits
generateda quasi-fiscaldeficitthat constitutedan importantsource of monetary
expansion. This problem can be examined in more detail by considering the
summarizedbalance sheet identityof the central bank in first differences:

C - H + NFL + NW (1)

Where C - central bank credits,H - base money (currencyplus reserves),

NFL = net foreign liabilitiesof the central bank and NW - the net worth of the
central bank. The dots indicate time derivatives. Note that the central bank
does not lend to the non-financialpublic sector. Note also that the variations
 11

In net foreign liabilitiesinclude a variation in foreign currencyand a capital

loss terms

whL
- NiL*S + NFL*E (2)

variable.
Where the star superscriptdenotesa foreignexchange-denominated
The variations in the central banks' net worth are determinedby the combination
of its income statement and revaluationaccounts

NW - i C - (i + E)NFLE - irR (3)

Where i., and ir are the interest rates on credits and bank reserves,
respectively. The interest rate on all foreign-denominated assets and
liabilitiesare assumed equal to i for simplicity. The hats indicatepercentage
changes6 . The interestrevenuesor expendituresof foreign-denominated
variables

are adjusted for the exchange rate depreciation. They can actually be broken
down between a cash and a capital loss terms7.
Combining equations (1) and (3) and solving for the variations in base
money yields:

H- C - NFL*E - icC + i*NFL*E + irR (4)

Equation (4) indicates that base money is expanded when there is an

increase in domestic credits or a decrease in net foreign liabilities,or when
the central bank runs an excess of interest expendituresover revenues. Note
that the capital gain/loss terms in equations (2) and (3) cancel out. A

depreciationgenerates a capital loss on the stock of net foreign liabilities

but has no immediate impact on base money creation. However, the capital loss
is eventually realized through interest payments or a decrease in the stock.

6 .
E - El/..
7 A

(i + E)NFL*E - i*NFL*E + NFL E, where the first term is the central

bank's cash expendituresand the second the capital loss on the stock resulting
from the depreciations.
 12
A much better insightcan be gained by stating equation (4) in real terms,
since the issue of real resource transfers becomes much more clear. Dividing
both sides of the equation by the domestic price levelp P, and the foreign-
denominated stocks by the internationalprice level, P *, and splitting the
nominal interest rates between the real rate and the inflation premium,
i - r + f, where r is the associatedreal rate and x the rate of inflation,one
obtains:

Hi/P irR/P cnfl*e - (rcc - r*nfl*e) (5)

8.
Where the lowercase letter indicate real stocks The term on the left is the
net seignorage revenulescollected by the central bank, defined as gross
seignorage minus the payment of interests on banks' reserves. The fact that
the real interest rate charged on domestic credits, rc, was always negative,
while the real cost of foreign liabilities,r*, was positive, implied a transfer
of seignorageto the recipientsof subsidizedcredits and the holders of foreign
liabilities. Such transfercan be verified by two alternativeways. First, by
examining the balance sheet of the central bank in real terms and, second, by
estimating the terms on the right hand side of equation (5).
Consider first the real balance sheet of the central bank in figure 4.
After 1982 the real stock of credits fell much faster than the real stock of base
money. The decline in real base money reflects the decline in the real demand
for financial assets, while the decline in real credits reflectsnot only this
factor, but also two additional factors, namely, the loss associated with the
real interest flows and the net repaymentsof foreign liabilities. This point
can be clarified further by rearrangingequation (5) and splitting seignorage
between the inflation tax, fh, and the real changes in the stock, hs

h + nfl*e + (r c - r nfl e) + irh (6)

Where the interests on reserves are disregarded for simplicity. Equation (6)

8
x - l/P.Note also that X/P - x + ¢x
 13
ehowe that real credits grow less than the real stock of base money when net
foreign liabilitiesdecrease or when the real interest flows are negative, and
grow more than real base money the larger the revenues from the inflation tax.
Of course, the two last terms capture the changes in the real net worth of the
central bank. The inflationtax is a real source of revenue to the central bank
and increases its real net worth, while the real interest losses decrease it.
When the inflation tax is smaller (larger)than the real interest losses there
is a decrease (increase)in the real net worth. A decrease in the real net worth
will be reflected in a decline of real credits relative to base money, unless
there is an increase in net foreign liabilities. If the real net worth
increases,but the if the inflationtax is also used to finance the net repayment
of foreign liabilities or the purchase of foreign assets, real credits also
decline relative to base money.

Figure 4

Balance Sheet of the Central Bank

400 . AI values (Base Yearclqe1O)

200- ^ TS

-100
-200 seMOE CREDTSFR NLIAD 8R

_eoo-.*...
-200

-300-

19892 1s3a 1964 19e5 199e 1967 1968 1969

From a different angle, if the central bank makes large net purchases of
foreign exchange from enterprises in the tradables sector in order to service
its foreign liabilities (interestand principal) or accumulate foreign assets,
 14
it is clear that credits will grow less than base money. Moreover, if the
central bank forgoes interestrevenuesby charginghighly negative real interest
rates on its credits, that imposes an additionalburden on its finances. The
smaller the flow of revenues the higher has to be the rate of monetary expansion
in order to achieve a certain credit target or to finance other operations.
That was essentiallythe situationfaced by the central bank of Yugoslavia
during the 1980s, althoughthe dominant factorsvaried from year to year. During
the mid-1980s the inflation tax was primarily used to finance the credit
subsidies and the real interests on foreign liabilities,although the central
bank also financed the net repayments of its foreign liabilities. Note that
during this period its real net worth declined. In 1988 and 1989 the decrease
in net foreign liabilities dominated the process of money creation. Note that
9.
in these last two years the real net worth increased
Table 4 provides further informationon the quasi-fiscaloperationsof the
central bank. The first section of the table shows gross and net seignorage
(gross seignoragewas measured by yearly changes in the nominal stock of base
money). The second section distributesseignorageamong the three major sources
of money expansion:the selectivecredits,the foreignexchange insurancescheme
and the central bank's operationswith the external sector. Each row shows the
10.
yearly changes in the stocks minus or plus the interest flows on the stocks
The figures are expressed as shares of GSP and are estimates, calculated from

9 In figure 4, credits and base money were divided by the December CPIs,
based in December 1980,while the foreignliabilitieswere converted into German
Marks (the dominant currency) at the cross-currencyrates of December 1980 and
multiplied by the Dinar/GermanMark rate of December 1980. The real net worth
was calculatedas a residual. This procedure is consistentwith equation (5),
and allows one to track the evolution of real foreign liabilitiesand the real
net worth net of the impact of the real devaluationsof the Dinar. If the net
foreign liabilitieshad been simply divided by the price leval, their increase
in 1983 would have been more pronounced, and the decline in real net worth
stronger.
10 For instance,in the case of selectivecredits the table show the changes
in selective credits minus the interest revenues on these credits, c - rcca
A similar procedure is applied in the case of the FX insurance scheme and the
operationswith the external sector.
 15

information on the stocks and interest rates. A detailed description of the

methodology used is provided in Rocha (1989).

Table 4
SeignorageRevenues on Base Money, 1981-88
(in X of GSP)
1980-84 1985 1986 1987 1988 1989

Gross Seignorage 2.7 3.6 4.2 4.9 5.5 12.1

Interests on Reserves 0.1 0.1 0.1 0.1 0.1 1.0
Net Seignorage 2.6 3.5 4.1 4.8 5.4 11.1
Selective Credits n.a 231 121 581 141 23X
FX Insurance Scheme n.a 501 451 50S 261 211
Net Liabilitieswith RoW n.a 27X 43% -81 60% 56X

Source: National Bank of Yugoslavia.

Table 4 shows that net seignorageincreasedthrough the 19809 to reach 11

percent of GSP in 1989. Regarding the main sources of monetary expansion,the
table also indicates the existence of two distinct periods. During the first
period (1985-87) the quasi-fiscal operations of the central bank were the
dominantsource of monetaryexpansion,with the foreignexchangeinsurancescheme
absorbing roughly 50 percent of seignorage. During the second period (1988-
89), therewas an effort to increaseinterestrates on domesticcredits to levels
closer to inflation. During this period, the large accumulation of foreign
assets (a decreasein net foreign liabilities)became the dominantsource of base
money creation. Thus, a large share of seignoragerevenueswas used to finance
the build-up of foreign reserves that preceded the stabilization program.
However, the volume of resources channelledto finance quasi-fiscaloperations
was also significant,especially in 1989, when it reached more than 4 percent
of GSP.1
Although the above analysis shows that central bank deficits were an

11 These estimates are based on data on stocks and interest rates and are
subject to some measurement errors. Thus, the objective of the exercise is to
provide information on trends and orders of magnitude. Nevertheless, the
findings are broadly in line with those provided by other authors (Bole and
Gaspari 1990 and Mates 1991), which also show the increasing burden of real
interestpayments in 1985-87,followedby a period (1988-89)where the decrease
in net foreign liabilitiesdominated the process of money creation.
 16

important source of monetary expansion and inflation, it might also suggest that
these deficits were only indirectly related to enterprise imbalances, and that
the link between domestic imbalances and inflationary finance was limited to
central bank operations. However, the whole volume of inflationary taxation was
not limited to the collection of seignorage on base money. Since commercial
banks paid negative real interest rates on all deposits through most of the
1980s, the total volume was much larger than that collected from base money

alone. This point must be stressed, since it brings to the fore the
distributional aspect of inflation in Yugoslavia, particularly the distribution
of real resources from holders of Dinar assets towards enterprises. Indeed, the
subsidies on domestic credits helped enterprises bear the burden of the external
transfer and kept a large number of loss making enterprises afloat during the
1980s. Furthermore, the dependency of these enterprises on subsidized financing
has always been a serious obstacle to successful stabilization in Yugoslavia.
An accurate calculation of the total resource transfers would require
detailed information on interest rates and the maturity breakdown of domestic
deposits and credits. The absence of critical pieces of information precludes
the conduct of such exercise. However, the potential size of seignorage
collected by the consolidated financial system can be appreciated by comparing
the relative magnitudes of gross seignorage on base money and MI in table 5.12

Table 5
Gross Seignorage on Base Money and Ml, 1980-89
(In % of GSP)

1980-84 1985 1986 1987 1988 1989

Base Money 2.7 3.6 4.2 4.9 5.5 12.1

Ml 4.8 5.0 9.1 7.7 11.0 24.5
Sources: National Bank of Yugoslavia and International Financial Statistics.

12 Measured by yearly changes in the respective stocks, as shares of GSP.

 17

Although these numbers may seem excessive, particularly in 1989, one has
to consider a number of factors. First, although the shift out of domestic money
was considerable (figure 5), the continuous increase in seignorage reflects in
part ongoing portfolio adjustment to an inflation that also accelerated

continuously. Second, two factors contributed to the unusually large seignorage

revenues in 1989. For one, the very high inflation in 1989 was to some extent

unexpected. Also, there was an unusually large expansion of credits and base
money in December of that year. Finally, in the case of M1 one has to consider

the fact that enterprises were also taxed, since they also held deposits. Thus,

although they were the main beneficiaries from the inflationary transfers, their
net gains were smaller than indicated in table 5.

Figure 5

Real M2, FX Deposits and M3

3 ~~~~~~~CGa9Q
. 1Qa5D

-7

19e1 1992 1993 1994 1995 1996 1997 1999 1999

o MZ COI flSr) + FX D0QCI ItS o 43

3. Exchange Rate. Wages and Prices.

The simultaneous occurrence of large real devaluations and increasing

inflation raise the question of whether there were other channels of transmission

of such exchange rate shocks. One popular model of inflation focuses on the
 18

combination of real exchange rate targets, rigid real wages and monetary
accommodation. It is actually a variant of a broad class of distributional
conflict models of inflation and has been labeled the Pazos-Simonsen mechanism

(e.g. Dornbusch 1986; Dornbusch and Simonsen 1987; and Pazos 1978). It is useful
to review a basic version of the model in order to assess its relevance to
Yugoslavia. To this end, assume that prices are determined by a mark-up over
costs, including labor costs and the costs of intermediate inputs, as in equation
(7):

Pt = aLWt + aMEtPMt (7)

Where P is the price level, W is the nominal wage, E is the nominal

exchange rate, PM* is the foreign price of the imported intermediate input, aL
and aM are the unit labor and intermediate goods requirements, respectively, and
the subscript t is the time period. The mark-up is set equal to zero for
simplicity. The real exchange rate is defined by:

et
et = EtPt*IPt
t t t(8 (8)

Where e is the real exchange rate and P is the foreign currency price of
the good competing with the country's exports in world markets. Substituting
(8) into (7) and assuming that all international prices are equal to unity, a
relationship between the real wage, w, and the real exchange rate is obtained:

wt= Wt/Pt = I/aL - (am/aL)et (9)

Figure 6 illustrates the relationship between the real exchange rate and
the real wage in Yugoslavia13 . Note the two periods when there was a clear
effort to maintain the real exchange rate at a targeted level: 1983 to mid-1986
and 1988 to mid-1989. In 1986-87 the real exchange rate target was partly
abandoned and real wages were allowed to increase. The same happened in the

13 The real exchange rate was measured by a simple basket of two real,
bilateral exchange rates (the Dinar/US$ and the Dinar/DM) with equal weights.
The average real wage was measured by the average nominal personal income divided
by the CPI.
 19

second half of 1989. However,also note that even in these periods the real wage
was usually below the levels of the 1970s.

Figure 6

Rea I Exchange Rate and Rea I Wages

180 indices (1980 = 100)

170 e

150 Peril Exchange Rate

140-

130 -
120-

110

100

SC)
70

1977 1978 1979 1980 1981 1982 1983 1984 198S 1986 1987 1998 1999 1990 1991

Of course, one has to bear in mind that unit labor costs decreased much
less than real wages. For one, the overall decline in labor productivityoffset
in part the fall in real wages. Also, the impositionof heavier payroll taxes
and other enterprise taxes also tended to increase unit labor costs. In the
enterpriseswhere the problem of excess labor was more severe, unit labor costs
incltding taxes actually increased. Nevertheless,the figure does reveal the
effort to drive down real wages and enforce a real devaluation.
The question is how the observed decrease in the real wage was obtained
in Yugoslavia. The Pazos-Simonsenmechanism postulates that, under real wage
rigidity, it takes an increase in inflation to engineer a decline in the real
 20
wage. To understand this mechanism,assume that the adjustmentof nominal wages
follows the rule Wt - cPt
1 1 , that is, a backward indexationformula. Clearly,

the parameter C is the peak real wage, that is, the level of the real wage
immediatelyafter the adjustment,while the effective real wage is wt - Wt/Pt.
It is easy to note that the extent to which the effective real wage falls below
the peak depends on the rate of inflationbetween t-l and t. The relationship
between the real exchangerate and the rate of inflation,ITt, follows by a simple
substitutions

c
wt - _ . 1/aL - (aM/aL)et (10)

The relevance of the Papoe-Simoneenmechanism for the Yugoslav case can

be assessed by examiningthe rules governing the adjustmentof nominal wages in
the 1980s. The basic principle that was followedin Yugoslavia in the 1980swas
the principle of relativeperformance. Any enterprisewas allowed to pay higher
than average wages if it could demonstrate that it had achieved a better than
average performance. This principle was formulated as an attempt to avoid
excessivewage payments and was made operationalthrough a formula of "business
performance" (Konovalov 1989; Saldanha 1989; and Vodopive,c1989). The
implications of this principle for the short-run behavior of real wages are
unclear, since there were no explicitrules for wage adjustmentswithin the year.
In 1987, however, the government introduced an explicit formula for
adjustmentswithin the year. It divided total wage payments for each enterprise
between two parts. The first part was considered as a basic labor cost, and
subject to periodic adjustmentswithin the year. The second part was determined
by the relative performance of the enterprise,as discussed above. The first
part was defined as 80 percent of enterprises'net operating income (revenues
minus non-labor costs) in the previousyear, adjusted for increases in the cost
of living--awage indexationrule.14

14 Bole and Gaspari (1990) also indicatethat indexationbecame widespread

after 1987.
 21
The impact of the introductionof wage indexationin 1987 may be examined
by inspecting the dynamic properties of the rate of inflation. Rewriting
equation (7), (8) and the backward indexationrule in terms of growth ratess

fft aLWt +
* ,MEt (11)

Et fft + et (12)

Wt "t-l (13)

Where aL is the share of labor in total costs (fL + aM - 1), the hate
indicate percentage changes and the foreign prices are again assumed to be
constant for simplicity. If the nominal exchangerate is adjusted according to
purchasing power parity, it follows thats

lft it-
1 (14)

Therefore,under these conditionsthe inflationprocess is fully inertial.

Of course, this crude version of the model assumes implicitlythe existence of
monetary accommodation. Indeed, the specificationof the aggregate demand and
the central bank's reaction function indicatethat full inertia only results if
there is also full monetary accommodationby the central bank (Cardoso1981 and
1983; Barbosa and McNellis 1989; and Barbosa and Vale 1989).15
To examine whether the data supports the hypothesis of fully inertial
inflation, the stochasticcomponent of the inflationaryprocess must still be
specified. One possibilityis simply to add a disturbanceterm to equation (14),
thus modelling the rate of inflation as a random walk:

Xt . Xt-i + Ut (15)

Where ut is a white noise stochasticprocess. A second possibility is to

model the rate of inflation as a local-levelmodel, i.e., a random welk plus
noise model (Harvey 1989). That assumes that inflation consistsof an inertial

15 The model can also be extended in other directions, such as the non-
synchronizationof wage adjustmentsacross differentclasses of workers, end the
declining periodicity of wage adjustments (Simonsen 1989).
 22

component, as in equation (15), plus a random noise (Barbosa and Vale 1989):

fft t +r ft (16)

ff t' t-1+t (17)

Where ff is the inertial component and et is a white noise stochastic

process. The reduced form of the local-levelmodel is an ARIMA (0,1,1) (Harvey
1989), while the first model is an ARIMA (0,1,0). In either case we will be
testing for the existence of unit roots in the inflation series. Table 6
reproducesthe results of the augmented Dickey-Fullertest (Dickey and Fuller,
1979 and 1981; Dickey et al 1986; and Nelson and Plosser 1982) for the monthly
rates of inflation (CPI) over successivesample periods,with the beginning of
the sample fixed at January 1977. The recursive estimates of the lagged
inflation coefficient are shown in figure 7.
The results indicatethat during 1988 the nature of the stochasticprocess
generating the inflation series changed. Until 1988 the hypothesis of a unit
root in the inflation series is rejectedat the 1 percent level of significance.
At the same time, the trend coefficient is quite significant. Therefore, the
inflation series seems to follow a trend-stationary(TS) process during this
period, as opposed to a differencestationary(DS) process (Nelson and Plosser,
1982)16. In sum, the stochastic nature of the inflation series during this
period does not suggest the existence of inertia.
When the sample is extendedbeyond 1988, however,the hypothesisof a unit
root and a DS cannot be rejected. Moreover, figure 7 indicates that the
structuralbreak started in June 1988, following the real devaluation of the
previous month and the maintenanceof the real exchange rate at the new target.
Note also that after the real devaluationof mid-1988, inflationrates stabilized
on a higher plateau,while real wages fell initially (figure 8). Such behavior

16 The statistic associatedwith the lagged rate of inflation is compared

against the Dickey-Fullertables for the r distribution (where the estimation
includes the constantand the trend variable). The significanceof the term in
lagged differencesis tested againsta standardt-distribution. See also Harvey
(1990).,
 23

Table 6
Augmented Dickey-FullerTests for the Monthly Rate of Inflation
Estimated EquationsD(dt) - aO + Olt + a 2 wt-1 + a 3 D(nt_.) + et
AAA A

Sample ao a1 82 a3
77.01-87.12 -0.01 0.0004 -0.73 -0.05
(-1.87) (5.32) (-6.61) (-0.57)
77.01-88.06 -0.01 0.0004 -0.82 -0.03
(-2.09) (5.73) (-6.76) (-0.29)
77.01-88.12 -0.01 0.0003 -0.52 -0.21
(-2.14) (4.57) (-5.44) (-2.50)
77.01-89.06 -0.01 0.0002 -0.18 -0.40
(-1.77) (2.69) (-2.22) (-4.70)
77.01-89.12 -0.01 0.0002 0.06 -0.54
(-1.0) (1.09) (1.20) (-6.56)
Notess D is the differenceoperator, D(wt) wt f It-i
Computed t-statisticein parertheses

Fi. gure .7

Recursive Estimates of the Lagged Inflation Coefficient

0.5 -

0.0

-0.5-\, ,/
I.,,'-. ;:'

-1.5 Mm" RN I filil M,

78 79 80 81 82 83 84 85 86 87 88 89
 24
is consistent with the Pazos-Simoneen mechanism, although inflation did
accelerate later on.17
Although the acceleration of inflation in 1989 iG partly associated with
the large build-uo of foreign reserves by the central bank, the large increase
in real wages also stands out as a major cause (figure 8)e Note also that such
increase in real wages was to a good extent forward looking--apreemptiveaction
against the expectationof a stabilizationprogram and a wage freeze (see Helpman
and Leiderman 1990, for a model of inflation based on forward-lookingwage
formation and monetary accommodation).

Figure 8

Monthly Inflation and the Real Wage, 1987-89

0.8

V 0.0
" 0.4

ito Inflation ;fo

100 , ,,- si
au.~~~~~~~~~~~~~~~V 0.
_ 90. 0.o

70 Rea Wage

1987 1988 1989

Two conclusions may be drawn from the analysis above. First, the second
experiment with real exchange rate targets had a much stronger impact on
inflation than the first, despite the smaller real devaluation. That reflects
the relativelylow downwardresistanceof real wages in the early 1980s, and the
increased real wage rigie ty at the end of the decade. Second, the lagged

17 The "blip" in the coefficientin late 1987 reflectsthe real devaluation

of that same period. Since that devaluationwas not sustained (figure6), the
coefficientdeclined during the first semester of 198C.
 25

inflation coefficient kept increasing in 1989, despite the fact that wage
formation shifted partly from backward to forward looking. Thus, although the
test is capturing the increased importance of these non-fiscal factors at the
end of the decade, the results also lend support to the observationmade by other
authors, namely, that such univariate tests are not sufficiently strong to
differentiatebackward-lookinginertia from a forward-lookingwage mechanism
(Helpman and Leiderman 1988 and 1989),18

4. Past StabilizationAttempts and the 1990 ProLram.

The failure to correct the internalimbalanceswas the main reason behind
the failure of various stabilizationattempts in the mid-1980s, vhich relied
mostly on wage-price controls. Particularlynoteworthy was the stabilization
attempt of mid-1988, which attempted to curb inflation through the imposition
of progressivelydeclining targets on the growth of money and wages, and the
attempt to impose financial discipline on enterprises through increases in
interestrates. To achieve that purpose,the authoritiesintroducedindexation
of time deposits above three months. No fiscal support to stabilizationwas
envisaged in the 1988 program.
The inconsistenciesof the mid-1988 program were aggravated by a real
devaluation(figure6) whose rationalewas unclear, since the countrywas already
running a large currentaccount surplusand was also engaged in debt rescheduling
negotiationswith private foreign banks. As it happened, the real devaluation
defeated one of the main purposes of the external debt rescheduling,which was
to relieve the pressure of external debt payments on the domestic economy.
The attempt to maintain the real exchangerate undervaluedin the context
of formal wage indexationresulted in a strong accelerationof wages and prices,
as shown above. The real devaluation also increased the burden of foreign
interest payments, thus offsettingin part the benefits of the reschedulingof
commercialdebt. The pressureto financehidden losses,particularlythe central

18 However,Petrovic (1990) formulatesand explicitmodel and concludesfor

the existence of inertia in the Yugoslav inflation.
 26
bank'o own deficit, led policy-makersto abandon the monetary targets soon after
their implementation. Faced with higher real interest rates on bank credits
several enterprisessimply stopped paying, aggravatingconsiderablythe already
severe problem of non-perfcsmingloans in the commercialbanks.
The failure of the 1988 stabilizationprogram showed the futility of
implementing another program without addressing the fundamental domestic
imbalances. Thus, during 1989, consensuswas reached about the need to generate
a surplus in the non-financialpublic sector in order to cover losses elsewhere
in the economy,even though there was less certaintyabout the requiredmagnitude
of the fiscal adjustment, or the best strategy to deal with the loss-makers.
In the case of the central bank's own deficit, the solutionwas clear, and
consisted basically in transferring the servicing of its foreign exchange
liabilities,and its credit subsidiesto favored sectors to the federal budget.
In the case of enterprisesand commercialbanks, the situationwas less clear.
Although enterprise losses where regularly calculated, the magnitude of
accounting problems decreased the reliance on the available figures. The
situation of commercialbanks was not fully transparenteither. The share of
non performing loans was known to be large, but the estimates were still
tentative. In addition,the governmenthad not completed the design of a well-
defined strategy to deal with the loss-makers. For instance, there were still
doubts of whether to provide a fiscal subsidy to lose-makers,while submitting
them to restructuringprograms (involvinglay-offs, debt write-offs, selective
improving investments,changes in management, and so on), or whether to let the
law of natural selectionoperate freely. In this case,
Darwinian-Schumpeterian
the number of bankruptcieswas expected to increasemuch more rapidly, and the
fiscal resources would be directed towards social programs, as opposed to loss
making enterprises. The final design of the program shows that policy-makers
opted for the second approach.
The stabilizationprogramof 1990 (actuallylaunchedin mid-December1989)
comprised a variety of measures in the areas of incomes, monetary, exchange,
trade, and fiscal policies (see Coricelli and Rocha 1991 for much greater
 27

detail). The program also included a 60-day tolerance limit for enterprise
arrears, as a device to force inefficient enterprises into bankruptcy. The
government could implement this measure due to the existence of a very
centralized system of payments in Yugoslavia, and it was announced as one
component of a future comprehensiveprogram of restructuringand privatization
of banks and enterprises.
Incomes policy comprised a 6-month freeze in the exchange rate, nominal
wages, and a set of public sector prices accountingto 20 percent of the CPI.
The exchange rate freeze was later extended to 1 year. The exchange rate and
public sector prices were frozen after a series of adjustments in late 1989.
However, the devaluationsthat preceded the freeze were rapidly eroded by the
increase in prices that followedthe implementationof the program, as shown in
figure 6.
Monetary policy consisted basically of a freeze in the nominal stock of
the central bank's net domestic assets, while allowing the central bank to
monetize foreign exchange inflows. The exchange and trade policy components
comprised a substantial relaxation of trade controls and the introductionof
currency convertibility.
Fiscal policy, a critical component of the program, consisted of an
expected fiscal adjustment of 5 percent of GSP designed to cover the central
bank's deficit, to support a banking restructuringprogram, to finance a social
program, and to cover some enterprisearrears. The envisaged increasein fiscal
revenueswas expected to come partly from exogenous increases in taxation (3.5
percent of GSP) and partly from the inverse Tanzi-Oliveraeffect (1.5 percent)
(see Olivera 1967; Tanzi 1977).
The stabilizationprogram of January 1990 achieved a sharp reduction in
the rate of inflation during the first semester of 1990 without recourse to
widespread price controls. As shown in figure 9, monthly rates of inflation
(retailprices) were reduced from 60 percent in December 1989 to almost zero in
May and June. A slowdownof economicactivity also followed the implementation
of the program, as indicatedby the 10 percent decline in industrialproduction
 28
during 1990. However, such a decline in activity had already started in the
second half of 1989 (figure 9), and cannot be blamed entirely on the
stabilization.
There are no indications that monetary policy was unduly restrictive in
the first semester. Indeed, a large repatriationof foreign assets held abroad
by exporting enterprises,and increasedworkers' remittancesresulted in a US$3
billion increase in reserves in the first semester. That, in turn, resulted in
a large increase in the real stocks of base money and Ml during the same period.
In addition, the December monetary and credit "blips" (figure 9) also suggest
that liquidity conditionswere not excessivelyrestrictiveat the start of the
program*
Even though monetarypolicy did not appear excessivelyrestrictive,a large
number of enterprises experienced severe difficulties in meeting their
obligationsduring the first semester,revealingtheir high degree of dependency
on bank credits. Out of 25,000 enterprises in the socialized sector, 7,000
enterprises had fallen behind their obligationsby at least one day, 3,000 had
accumulatedarrears for 30 days and 350 were declaredbankrupt9for having failed
to meet payments for 60 days. A large number of enterprises interruptedwage
payments in May, in order to postpone bankruptcy. The situation at the end of
the first semester was very differentiated,with one group of enterprises
increasing wages above the ceiling, and another group unable to make wage
payments within the ceiling.
Pressures to relax monetary and credit policiesmounted during the first
semester,leading effectivelyto a relaxationat the end of the semester. Bank
credits started growing very strongly after June, leading to further increases
in wages and finally to an increasein the rate of inflation to levels above 8
percent in September and October (figure9). The relaxationof monetary policy
also aborted the expected shake-out of the industrialsector, by keeping loss
making enterprises afloat, while also allowing them to resume wage payments.
Concern over the revival of inflation led the central bank to shift back
to a restrictivemonetary policy in early October. However, pressed by
enterprises 'o continue lending, the banks avoided a contraction of credit by
not complyingwith reserve requirements.The last indication of the difficulties
 3
0 ~~~~~~m
x~~~~~~~~~~~~~~~~~~~~~~~~~~~~

a~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i

- I I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~r

q . . . . . .. . . . . .
g F~~~~~~~~~~~~~~~~~~~~~~~~~~

CD
ft
 30

the central bank faced in conductingmonetary policy was a bizarre episode in

December 1990, when one of the regional branches of the central bank increased
credits by US$1.8billion equivalent of dinars, without prior notificationto
the board of governors, in order to finance lose making enterprises,pensions,
and agriculturalsubsidies (The Economist,January 12, 1991).
The expansion in central bank credits did not result in a monetary
expansionbecause of the large loss of foreignreserves that happenedin the same
period (more than US$ 3 billion in the last quarter). In fact, the situation
in the last quarter was exactly the reverse of the first quarter, involving an
expansion of domestic credits,a fall of foreign reserves and a nearly constant
stock of M1 (figure 9). Althoughthe open currentand capital accountsprevented
a further acceleration of inflation at the end of 1990, these episodes of
monetary decontrol and the large loss of reserves raised obvious doubts about
the sustainabilityof the stabilizationprogram.
The recognitionthat the exchangerate had become severelyappreciatedled
to a corrective 30 percent devaluation on January 1, 1991, as well as a the
suspensionof convertibility. These measureswere unavoidable,given the large
loss of reserves at the end oE 1990. However, they also led to an increase in
inflationin early 199L (5 and 10 percent in January and February, respectively),
despite the reimpositionof wage controls and efforts to reassert control over
monetary policy. Moreover, the erosion of the first devaluationled to a second
corrective devaluation of 45 percent in April 1991. Therefore, the challenge
faced by policy-makersin 1991 is to obtain the countrywideconsensus that will
allow the continuousenforcementof wage, monetary and fiscal policieswhich is
required to prevent the reemergence of an exchange rate-wages-pricesspiral.

5. Conclusions
The paper had two main objectives. First, to examine the main causes of
the accelerationof inflation in Yugoslavia during the 1980s and, second, to
review past and current attempts at stabilization. The paper showed that
inflation in Yugoslavia shares common elements with inflation in other high
indebtedcountries,despite initial appearancespointing otherwise. The common
elements are a large transfer of resources abroad not matched by an internal
 31

adjustment, and resulting in a large internalredistributionof real resources

through inflation. Yugoslavia seems to differ from other cases due to the lack
of transparencyof internalconditions. Instead of an open fiscal deficit,there
were complex interactionsamong enterprises,commercial banks and the central
bank, involving, among other things, the absorption and servicing of a large
stock of foreign exchange liabilitiesby the latter. The paper also showed how
other factors contributed to the sharp accelerationof inflation at the end of
the decade. More specifically,a large real devaluation in mid-1988 in the
context of an indexed economy drove inflationto a much higher plateau. During
1989, a preemptiveexplosion of real wages became a major factor of acceleration
of inflation.
The paper argued that the failure to correct the hidden losses in the
econory was the main cause of the failure of the various attempts at
stabilizationin the 1980s. The 1990 programvas the first program to recognize
the existence of hidden losses and the need for a fiscal correction,although
it also introducedother elementsto cope with inflationaryinertia. The program
the central bank's own deficit,and had initial success
succeededin elimitnating
in halting inflation. However, it became clear during the course of 1990 that
other losses had not been removed. Pressures to finance enterprisesand avoid
a liquidity crisis in the financialsystem resulted in a relaxationof monetary
policy in mid-1990 and a revival of inflationary pressures in the second
semester. Attempts to reimposemonetary control met considerabledifficulties
at the end of the year, includinga bizarre episode of expansion of central bank
credits withoutapproval by the board of governors.
It became also clear that the fiscal component was not consistent with
other elements of the program. In particular, it was clearly insufficientto
finance a social program of the magnitude that would be required,had loss making
enterprisesreally been forced into bankruptcy, and also to cover the needs of
the banking restructuringprogram. Seen from this angle, the Yugoslav program
of 1990 resembles other heterodoxprograms that had initial success in reducing
inflation,but later faltereddue to the insufficiencyof the fiscal adjustment.
At the same time, the events in the second half of 1990 also indicatethat,
for a stabilizationprogram to succeed in Yugoslavia,there must be much greater
 32
political resolve in dealing vith wage indisciplineand loss making enterprises
than vas observed during the course of 1990. It also brings the question of
whether financial discipline can really be imposed in the system only at the
macroeconomic level, and without the introduction of private ownership of
capital. This finally brin8s us the related question of whether stabilization
in Yugoslavia can ultimately succeed in the absence of a comprehensive
privatizationprogram.
 33

REFERENCES

Barbosa, F. and P. McNellis (1989). "Indexationand InflationaryInertia:

Brazil 1964-85". World Bank Economic Review, Vol. 3, (3); pp 339-358.
Barbosa, F. and Vale, P. (1989). " 0 Insucesso do Plano Cruzado: A
Evid8ncia Empiricade Infla9ao100% Inercialpara o caso do Brazil".In Barbosa,
F. and Simonsen, M., eds, Plano Cruzado: In6rcia x In62cia. Rio de Janeiro:
Editora Globo.
Bole, V. and Gaspari,M. (1990). "The Yugoslav "Way" to Hyperinflation".
Belgrade. Processed.
Cardoso, E. (1981). "Indexation,Monetary Accommodationand Inflation in
Brazil". Boston University. Center for Latin American Development Studies.
Dicussion Paper No. 49.
Cardoso,E. (1983)."Indexagaoe AcomodagaoMonet6ria:Um Teste do Processo
Inflacion6rioBrasileiro". Revista Brasileirade Economia, 37, (March);pp 3-
11.
Coricelli, F. and Rocha, R. (1991). "StabilizationPrograms in Eastern
Europe: A Comparative Analysis of the Polish and Yugoslav Programs of 1990".
Washington D.C., The World Bank. Processed.
Dickey, D., Bell, W. and Miller, R. (1986). "Unit Roots in Time Series
Models: Tests and Implications". American Statistician,40, pp 12-26.
Dickey, D. and Fuller, W. (1979). "Distributionof the Estimators for
Autoregressive Time Series with a Unit Root". Journal of the American
StatisticalAssociation,74 (June): pp 427-31.
Dickey, D. and Fuller, W. (1981). "Likelihood Ratio Statistics for
AutoregressiveTime Series with a Unit Root". Econometrica,49 (July):
pp 1057-72.
Ding, W. (1990). "Yugoslavia:Review of Public Sector Finance, Issues and
Priorities for Reform". Washington D.C., The World Bank. Processed.
Dornbusch, R. (1987). "Lessons from the German Inflstion Experienceof
the 1920s". In Dornbusch,R and Fischer, S. (eds), Essays in Honor of Franco
Modigliani. Cambridge,MA. The MIT Press. pp. 337-366.
Dornbusch, R and Fischer, S. (1986). "StoppingHyperinflation:Past and
Present". WeltwirtschaftlichesArchiv, 122 (1) (April):pp 1-47.
Dornbusch,R and Simonsen,M. (1988). "InflationStabilizationsThe Role
of Incomes Policy and of Monetization". In Dornbusch, R. Exchanae Rates and
Inflation. The MIT Press. pp. 439-466.
Gaspari,M. (1988). "FinancialCrisis in YugoslaviaSince the Early 1980s:
Causes and Consequences". Paper presentedat the Conferenceon FinancialReform
in Socialist Countries, Florence, October 1987.
Gordon, R. (1984). "Unemploymentand Potential Output in the 1980s".
Brookings Papers in Economic Activity, 15 (2): pp 537-64.
Harvey, A. (1989). ForecastinM.Structural Time Series Models and the
Kalman Filter. Cambridge,U.K., Cambridge University Press.
 .L~J

34

Harvey, A. (1990). The EconometricAnalysis of Time Series. Cambridge,

MA. The MIT Press.
Helpman, E. and L. Laiderman. (1988), "Stabilizationin High Inflation
Countriese Analytical Foundations and Recent Experience". Carnegie-Rochester
Conference Series on Public Policy, 28, pp. 9-84.
Helpman, B. and L. Leiderman. (1990). "Real Wages, Monetary Accommodation
and Inflation". EuroRean Economic Review, Vol 34, pp. 897-911.
Knight, P. (1984). "Financial Discipline and Structural Adjustment in
Yugoslavia". Staff Working Paper No 705. WashingtonD.C., The World Bank.
Konovalov, V. (1989). "Yugoslav Industry: Structure, Performance and
Conduct". Washington D.C., The World Bank. Processed.
Liviatan, N. and Piterman, S. (1986). "AcceleratingInflation and Balance
of Payments Crises, 1973-84". In Yoram Ben-Porath, ed. The Israel, Economy.
Cambridge,MA. Harvard University Press.
Mates, N. (1987). "Some Specific Features of Inflation in a Heavily-
IndebtedSocialistCountry". EconomicAnalysisand Workers'Management,21 (4):
pp 419-32.
Mates, N. (1990)."Inflationin Yugoslavia:SpecificForm of Public Deficit
Caused by Parafiscal Operationsof the Central Bank". Zagreb. Processed.
Mates, N. (1991). "Fiscal Reform Issues in Yugoslavia". Zagreb.
Processed.
Mencinger,J. (1987)."Accelerationof Inflationinto Hyperinflation: The
Yugoslav Experience in the 1980s". EconomicAnalysis and Workers' Manasement,
21 (4): pp 399-418.
Mencinger,J. (1989)."PrivrednaReforma i Nezaposlenost"(EconomicReform
and Unemployment).Privredna Kretanva Jugoslavive,March.
Montiel, P. (1989). "Empirical Analysis of High-Inflation Episodes in
Argentina, Brazil and Israel". IMF Staff Papers, 36 (September).
Nelson, C. and Plosser, C. (1982). "Trends and Random Walks in
MacroeconomicTime Series: Some Evidenceand Implications". Journal of Monetary
Economics 10 (2): pp 139-62.

Okun, A. (1970). "PotentialGNP, Its Measurement and Significance". In

Okun, A., The Political Economy of Prosperity. Washington D.C.,
Brookings Institution.
Olivera, J. (1967). "Money,Prices and Fiscal Lags: A Note on the Dynamics
of Inflation". Quarterly Review, Banca Nazionale del Lavoro, 20 (82): pp 258-
67.
Pazos, F. (1978). Chronic Inflation in Latin America. New York: Praeger.
Petrovic, Pavle (1990). "Inflationary Inertia in the Yugoslav
Hyperinflation:Some EconometricEstimates". Universityof Belgrade. Processed.
Rocha, R. (1989). "Inflation and Structural Adjustment in Yugoslavia".
Washington D.C., The World Bank. Processed.
Saldanha, FP. (1989). "Self-Management:Theory and Yugoslav Practice".
Washington D.C., The World Bank. Processed.
Simonsen, M. (1989). "Inercia Inflaciongriae Inflagao Inercial". In
Barbosa, F. and Simonsen, M., eds, Plano Cruzado: In6rcia x In6pcia. Rio de
 35

Janeiros Editora Globo.

Tanzi, V. (1977). "Inflation,Lags in Tax Collection,and the Real Value
of Tax Revenue". IMP Staff PaDers, 24 (March):pp 154-67.
Vodopivec, H. (1989). "Productivity Effects of Redistribution in a
Socialist Economy: The Case of Yugoslavia". Unpublished PhD dissertation,
Universityof Maryland.
 PREWorkingPaperSeries

Contact
IJ1Q Author for pa,per

WPS730 Wageand EmploymentPoliciesin Luis A. Riveros July 1991 V. Charles

Czechoslovakia 33651

WPS731 EfficiencyWageTheory,Labor Luis A. Riveros July 1991 V. Charles

Markets,and Adjustment LawrenceBouton 33651

WPS732 StabilizationProgramsin Eastern FabrizioCoricelli July 1991 R. Martin

Europe: A ComparativeAnalysisof Robertode RezendeRocha 39065
the Polish and YugoslavPrograms
of 1990

WPS733 The ConsultingProfessionin SyedS. Kirmani July 1991 INUDR

DevelopingCountries: A Strategyfor WarrenC. Baum 33758
Development

WPS734 CurricularContent,Educational Aaron Benavot July 1991 C. Cristobal

Expansion,and EconomicGrowth 33640

WPS735 TraditionalMedicinein Sub-Saharan JocelynDeJong July 1991 0. Nadora

Africa: Its Importanceand Potential 31091
Policy Options

WPS736 Wagesand Employmentin the SimonCommander August 1991 0. Del Cid

Transitionto a MarketEconomy FabrizioCoricelli 39050
Karsten Staehr

WPS737 External Shocks,Adjustment DelfinS. Go August 1991 A. Bhalla

Policies,and Investment: Illustrations 37699
from a Forward-looking
CGEModel
of the Philippines

WPS738 Tax Competitionand Tax RaviKanbur August 1991 J. Sweeney

Coordination:WhenCountries MichaelKeen 31021
Differ in Size

WPS739 ManagingFinancialRisksin Papua JonathanR. Coleman August 1991 J. Carroll

NewGuirsa: An OptimalExternal YingQian 33715
Debt Portfolio

WPS740 The OnchocerciasisControlProgram BernhardH. Liese August 1991 0. Nadora

in WestAfrica: A Long-term John Wilson 31091
Commitmentto Success BruceBenton
DouglasMarr

WPS741 WhenDoes Rent-SeekingAugment DavidTarr August 1991 D. Ballantyne

the Benefitsof Priceand TradeReform 37947
on RationedCommodities?Estimates
for Automobilesand ColorTelevisions
in Poland
 PRE WorkingPaperSeries

Contact
1]Q ~~~~~~~~~Author W ape
fr

WPS742 The Cost of the DistrictHospital: A. J. Mills August 1991 0. Nadora

A CaseStudyfrom Malawi 31091

WPS743 AntidumpingEnforcementin the AngelikaEymann August 1991 N. Artis

EuropeanCommunity Ludger Schuknecht 37947

WPS744 StainlessSteelin Sweden:Anti- GunnarFors August 1991 N. Artis

dumpingAttacksGood International 37947
Citizenship

WPS745 The Meaningof "Unfairin U.S. J. MichaelFinger August 1991 N. Artis

ImportPolicy 37947

WPS746 The Impactof Regulationon DimitriVittas August 1991 W. Pitayatonakarn

FinancialIntermediation 37666

WPS747 Credit Policiesin Japan and Korea: DimitriVittas August 1991 W. Pitayatonakarn
A Reviewof the Literature 37666

WPS748 EuropeanTrade PatternsAfter the Oleh Havrylyshyn August 1991 N. Castillo

Transition Lant Pritchett 37947

WPS749 HedgingCommodityPriceRisksin Stijn Claessens August 1991 S. Lipscomb

PapuaNewGuinea JonathanColeman 33718

WPS750 Reformingand PrivatizingPoland's Esra Bennathan August 1991 B. Gregory

RoadFreightIndustry JeffreyGutman 33744
LouisThompson

WPS751 A Consumption-Based DirectTax CharlesE. McLure,Jr. August 1991 CECSE

for Countriesin Transitionfrom 37188
Socialism

WPS752 Inflationand Stabilizationin Robertode Rezende August 1991 L. Ly

Yugoslavia Rocha 37352