See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/27191933

OLIVERA-TANZI EFFECT: THEORY AND ITS

MANIFESTATION IN THE CROATIAN STABILIZATION
PROGRAMME

Article · January 1996

Source: OAI

CITATIONS READS
4 1,568

2 authors, including:

Sandra Svaljek
Ekonomski institut, Zagreb
27 PUBLICATIONS   57 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Article on territorial fragmentation in Croatia View project

Economic growth View project

All content following this page was uploaded by Sandra Svaljek on 31 October 2017.

The user has requested enhancement of the downloaded file.

 OLIVERA-TANZI EFFECT: THEORY
AND IT'S MANIFESTATION IN THE
CROATIAN STABILIZATION PROGRAMME

Zoran Anušić and Sandra Švaljek

1. INTRODUCTION

Inflation is often considered to be the apple of paradise, despite all the

unpleasant experiences with it.

When the government consumption, according to some rough

estimations, amounts over 50°/o of gross domestic product, and when,
therefore, the changes in the field of public finances can have serious
implications, the possible effects of inflation on the budget become of
great concern, as well as the effects of inflation on the economy as a
whole. It is already well known that the power to issue money provides
some additional resources for the government, in the form of the so
called seigniorage. But, on the other hand, printing money may cause
inflation which could influence the real collection of "normal" taxes.
Recently, The Republic of Croatia has experienced high rates of
inflation accompanied by a decline in real public revenues. In other
words, Olivera-Tanzi effect occurred.

The aim of the paper is to explain the mechanisms of functioning of

Olivera-Tanzi effect, to describe the experiences of those countries
where that effect happened, to identify the circumstances under which
it arises, and to estimate the intensity as well as the reasons for its
occurrence in Croatia before and after the beginning of implementation
of the Stabilisation programme. The authors' intention was to find the
answer to the question whether the circumstances changed in last two
years, or the same things would have happened if the initial price
increase had been allowed. The authors want to induce further
investigations in the field of interrelations between inflation and fiscal

73
policy, and the investigation of relationship between the fiscal and
monetary policies as well.

In the second part of the paper the logic of the Olivera-Tanzi effect is
presented, taking the Olivera-Tanzi effect as a dynamic process
happening to the real revenue from inflation as well as to the real
revenue from "normal" taxes, in the period of accelerated inflation.

The third part is the overview of the experiences of a number of

countries that faced the selfperpetuating interrelation between inflation
and fiscal deficit. Olivera-Tanzi effect in the Republic of Croatia in the
period 1992 - 1994 will be empirically analysed in the fourth part. In
the fifth, concluding part we give some suggestions that follow from the
findings based on the Croatian experience, as well as the suggestions
for the future policymaking.

2. THEORETICAL FRAMEWORK OF THE OLIVERA-TANZI EFFECT

The theory has already become familiar with the term Olivera-Tanzi
effect and it has been used in describing the decrease in the real
government revenues in the period of rising inflation.1 Government
revenues could be considered a sum of revenues from "normal" taxes,
and the revenues from inflation tax. In connection with this, there are
two parallel processes going on behind Olivera-Tanzi effect. The first
one is the process of the reaction of the real inflation tax revenues to
the rate of inflation, and the second one is the reaction of the real tax
revenues from "normal" sources (e.g. sales tax, corporation income tax,
personal income tax) to the rate of inflation.

1
This phenomenon was first noticed after the very first hyperinflation (according to
Cagan's definition), the one that occurred in Germany after the World War I
(Bresciani-Turroni, 1936, in: Canavese, Heymann, 1992). The phenomenon got its
name later, after J. H. G. Olivera (1967) and V. Tanzi (1977) who were the first to
interpret it by means of the standard analytical tools.

74
The revenue from the inflation tax can be seen as the tax rate - the
inflation rate, B multiplied by the tax base - the real money balances m.
According to Cagan (1956) the real demand for money changes with
respect to the rate of inflation. In other words, the real demand for
money moves according to the expression:

(1) m=m0 e-aB,

where a denotes the coefficient of the sensitivity of the real demand for
money with respect to the anticipated rate of inflation, and mo denotes
the demand for money when n = 0. The real revenue from the inflation
tax is, then,

(2) I R = B m0 e-aB

The inflation tax maximizing rate of inflation is B=1/", that means that
the revenue from the inflation tax reaches its maximum when the
elasticity of the demand for money with respect to the rate of price
change is unity.2

On the other hand, the revenue from "normal" taxes also responds to
the price increase. If the tax system is generally progressive, the real
tax revenues could stay unchanged when prices increase, or even rise.
Nevertheless, if the inflation is high, it can be expected with greater
certainty that the government would collect smaller amount of taxes in
real terms, and that the loss of real resources would be the greater the
longer the lag between the moment the taxable event occurred and the
moment the government collects the tax (Tanzi, 1978). The case of the

δ IR = m 1
e − e
−

δ 0 =0 when =

m, = m0 e −
=
m0 e −

75
loss of real tax revenues induced by inflation is known as Olivera-Tanzi
effect.

The real value of the "normal" taxes the government collects, assuming
that the elasticity of the real tax revenues to the increase in real GDP,
is then:

T0
(3) T = ,
n
1+

that is

T = T0 e
−
(4) ,

where T0 denotes the effective tax burden in the inital period, n is the
average collection lag expressed in months, and β denotes the
coefficient of the average collection lag expressed in months.

Total revenue in the period of inflation will then be the sum of the
expressions (2) and (4), or

TR = m0 e + T0 e
−
(5) −
.
3
The rate of inflation that maximizes the total government revenue is
1
(6) *=
T *
+
IP *

T *
The expression in the denominator is positive, which
IP *

δ TR δT δ IR 1
= + = 0, when *=
δ δ δ T *
+
IP *

76
means that the rate of inflation maximizing total government revenue
is, due to the destructive influence of the inflation on the real revenues
from "nominal" taxes, lower than the rate of inflation that would
maximize the government revenue in the absence of that influence. In
other words,
1 1
<
T *
+
IP *

Assuming the rate of inflation were exogenous, the choice of the

adequate rate of inflation involves great responsibility, for it can
influence the revenues from the inflation tax on one hand, and the
revenues from the "normal" taxes on the other hand. The above
analysis shows that:
a) the lower the initial level of the real money balances, and the
greater the share of the government revenues collected
through "normal" taxes, the greater the fall of the total
revenues caused by inflation, and the lower the rate of inflation
maximizing the total revenues,
b) the lower the initial tax burden, and the bigger the share of
inflation tax in the total revenues, the higher the rate of
inflation that maximizes the total revenues,
c) the greater the sensitivity of demand for money with respect to
price changes, the smaller the revenues from inflation tax, and
the lower the rate of inflation that maximizes the total
revenues,
d) the larger the collection lag, the bigger the erosion of the real
revenues caused by inflation, and the lower the rate of inflation
that maximizes the total government revenues.4

4
It has to be pointed out that the former analysis of the revenues from the inflation
tax is based on following assumptions: 1) the only cause of inflation is the
inflationary finance of the fiscal deficit, 2) the money supply changes only as a
consequence of the deficit financing, 3) that inflationary expectations equal the
actual price changes, 4) that the monetary multiplier equals 1, that is, private banks
do not evolve in money creation.

77
Olivera-Tanzi effect is the case when the price increase exceeds the
optimal rate of inflation and when the erosion of real collection of tax
revenues surmounts the gains from any further money creation. The
moment when this will happen depends on parameters m0, T0, β and ".
On the other hand, whether the Olivera-Tanzi effect would operate and
will it operate strongly or not, will depend on the features of the tax
system. In other words, Olivera-Tanzi effect will be the' stronger,
a) the lower the progressivity of all forms of taxes, that means the
lower the elasticities of "normal" tax revenues with respect to
the price increase,
b) the higher the share of indirect sales taxes and excises with
specific tax rates in total tax revenues,
c) the more difficult the tax system accommodates to the price
increases by introducing discretionary measures like e. g.
increase in tax rates, shortening the statutory lag for paying
taxes, by raising up penalties for paying taxes after they fall
due, indexing tax bases and strengthening the tax
administration.

Some analyses show that the Olivera-Tanzi effect causes the increase
in fiscal deficit. This follows from the fact that the nominal government
expenditures increase almost instantaneously and in the same
proportion as prices, that is, they do not change in real terms. Along
with the decline of the real government revenues, this causes the
steady rise of the fiscal deficit. Therefore, having no possibility for
revenues to keep the pace with the price increase and with the rise of
nominal expenditure other then the initial deficit, the government faces
certain amount of unintended deficit, which depends on the rate of
inflation.

The inflation itself, however, is often a consequence of unsustainable

fiscal deficit. In other words, when there are limited possibilities for
broadening the tax base, when the domestic capital market is
insufficiently developed, and the level of external reserves is low,
money creation is the only possible way to finance the surplus of

78
government expenditures over the government revenues. The mutual
relationship between the fiscal deficit and inflation is almost perfect
under such circumstances. It is then obvious that the simultaneous link
between the inflation and the fiscal deficit causes a selfperpetuating
interference of those two phenomena, whereby every further step of
this interference means a higher rate of inflation and deficit. It seams
logical to raise the question whether there is some rate of inflation at
which those tendencies stop, that means some rate of inflation that
would provide for the government some inflation tax revenue that
would be sufficient to cover the initial deficit as well as the deficit
provoked by inflation, or does that process necessarily lead to an
explosion of the deficit and inflation (Kiguel, 1989; Choudhry, 1990;
Canavese, Heymann, 1992).

If the initial amount of fiscal deficit is

(7) D0 = G0 – T0 ,

in times of inflation, owing to the collection lag it becomes

(8) DB – G0 - TB ,

where

(9) DB = D0 + d (B) ,

In the above expressions D = fiscal deficit, T = tax revenues, G =

government expenditures, d = a part of the actual deficit, which
depends on the rate of inflation.

Equilibrium is reached when the real inflation revenues match the total
deficit, that is when DB = IR (Choudhry, 1990).

The fiscal deficit function is the increasing one since its first order
derivative is positive. That means that the real fiscal deficit increases

79
together with the rate of inflation, and that it reaches its maximum
since its second order derivative is negatives.5

If one assumes the function of the demand for real money balances of
the Cagan type, and the shape of the fiscal deficit function mentioned
above, the equilibrium inflation rate can be determined graphically
(Figure 1).

The equilibrium is reached where the two schedules intersect. There

are two such points - one at the low inflation rate, B L, and one at high
rate of inflation, B H. Since the second order derivatives of both

δT
= D >0 ,
δ
2
δD
=- 2
D <0,
δ 2

80
schedules are negative in the neighbourhood of low equilibrium
inflation rate, stable equilibrium is reached at that inflation rate.
Namely, at that level of inflation any further monetary expansion leads
to the increase in real revenues from the inflation tax and diminishes
the deficit, thereby reducing the need for further inflationary finance
and leading the system to the equilibrium. If the initial fiscal deficit is so
high that there is no monetary expansion that could cover it, there is no
equilibrium rate of inflation. Such unsustainable level of fiscal deficit
could be even the one that is lower than the maximum revenue from
inflation, since there is the part of unintended fiscal deficit caused by
the rate of inflation maximizing the inflation tax. When there is no
stable equilibrium of that dynamic process, any further attempt to
increase inflation revenue through money creation deepens the gap
between the real government revenues and the real government
expenditures (Choudhry, 1990).

From the above analysis one can conclude that if there is a possibility
for Olivera-Tanzi effect to occur, there are at least two arguments in
favour of low, rather then high inflation. High inflation can, on one
hand, lead to great loss in real tax revenues. On the other hand, high
inflation causes the increase in the difference between the real
government revenues and real government expenditures and can
provoke the selfgenerating and intensifying interrelation of fiscal deficit
and inflation, thus destabilizing monetary and fiscal sphere of
macroeconomics.

3. OLIVERA-TANZI EFFECT IN SOME

EMPIRICAL INVESTIGATIONS

Many investigations imply that the Olivera-Tanzi effect followed almost

all high inflations. Among those investigations there are those on the
phenomena at the time and after extreme inflationary financing in
Austria, Germany, Poland and Hungary after the World War I (Franco,
1990), in Columbia, Dominican Republic, Brazil and Indonesia in the
late 50's and 60's (Aghevli, Khan, 1977 and 1978), in Argentina in the

81
late 60's and 70's (Tanzi, 1977) and in Bolivia in the period of
hyperinflation in 1985 (Sacks, 1985).

The experiences of Austria, Hungary, Poland

and Germany after the World War I

The fall of the Habsburg monarchy caused economic crises in two

newly established, and therefore politically and economically rather
weak countries - Austria and Hungary. As it seems natural for such
political circumstances, these crises were reflected in high budget
deficit of both countries, which resulted in the first hyperinflation to be
recorded. The stabilization programmes in both countries were initiated
and supported by the League of Nations, which was the predecessor of
to-day IMF. The most important parts of the stabilization programmes
were fiscal reforms, and the stopping of the decrease of the value of
domestic currency (Franco, 1990). The data show that in Austria, there
were no significant changes in real government expenditures, while at
the same time there was a sudden increase in tax revenues, which
made it possible to balance the state budget. It is very interesting that
the League of Nation, predicted the similar developments in Hungary,
drawing upon the experiences of Austria. The predictions happened to
be correct. In Hungary there also occurred the decline of the fiscal
deficit during the implementation of the stabilization programme, that is
after stopping inflation, whereby the major contribution to that
decrease came from the rise in exactly those taxes that were
untouched by the fiscal reform (consumption tax, duties). The League
of Nations cited in its final reports on the stabilization programmes in
Austria and Hungary that "budget equilibrium followed (the
stabilisation), it did not precede".

The Polish crisis in the beginning of the 20's of this century was caused
by the political factors, too. In 1919, after having been split among
Russian, Habsburg and German empires for hundred years, the new
Poland appeared within entirely new borders. The Polish economic crisis

82
was characterized by the growth of the public sector and very high
discrepancy between the public expenditures and public revenues. In a
young country, that has only a limited choice of instruments for
financing fiscal deficit, inflation was a logical consequence. The
stabilization programme that started in 1923 did not succeed in
stopping the expansion of the public expenditures, and the inflationary
finance partly continued even after the central bank became
independent from the state. Namely, the government retained the right
to print coins and small notes. In the 1924 and 1925 the money
creation decreased, and due to the positive influence of the fall of
inflation to the real tax revenue, the budget was almost balanced.

The case of Germany is similar to the cases of Austria and Hungary

regarding the reactions of real tax revenues to the rate of inflation. On
the other hand, the difference is in that the explosion of prices in
Germany after the World War I. was not simply the consequence of
insufficient power of the state to collect resources to finance public
needs, but it was the result of the collapse of the public finances
caused by paying war reparation obligations. The stabilization
programme aimed at placing the existing tax system on a "golden
basis" in the period of the high inflation (in 1919 and 1920 it reached
about 900% a year) and very low coverage of the public expenditures
by the public revenues. Though there was no lowering of public
expenditures, the situation in the budget recovered very quickly due to
the increase in real collection of tax revenues.

In all the four examples of hyperinflation there occurred losses in the

fiscal system, that is, there was Olivera-Tanzi effect. The main
prerequisite of the Olivera-Tanzi effect in the cases mentioned above
was inflationary financing of the budget deficit that resulted in
extremely 'high rates of inflation, which, in the situation of inefficient
tax administrations of those countries, and together with the long
collection lags caused inflation to "bite" a part of the tax revenues.

83
The experiences of Brazil, Columbia, Thailand,
Dominican Republic and Indonesia

The following cases of the countries for which the relationship between
the fiscal deficit and inflation was to a certain extent different. That are
the cases of Dominican Republic, Columbia and Thailand in the period
1961 - 1974, and in Brazil between 1964 and 1974. Those countries
had in that period the rates of inflation of various levels, but their
common characteristics were undeveloped capital market and the
limited space for broadening the tax base. Therefore the inflation tax
remained as the only possibility of financing even the lowest budget
deficit. The proof that there was Olivera-Tanzi effect can be drawn from
the fact reached by Aghevli and Khan (1978), which states that in all of
the four countries concerned it took much more time for tax revenues
to accommodate the price changes than it was the case with public
expenditures, regardless what was the initial inflation level.

A typical example of the Olivera-Tanzi effect were the events in

Indonesia in the period 1951 - 1972 (Aghevli, Khan, 1977). The price
increase started at the end of 50's when the financial needs of the
government rose due to the political turmoil, and the crisis was even
sharpened by heavy draught that in 1961 led to the drastical decrease
in food supplies and to the additional price increase. That caused the
serious increase in nominal government expenditures. As the rise in the
government revenues considerably fell behind the price increase, the
government was forced to finance the deficit by money creation. The
selfperpetuating process of increase in deficit and inflation occurred, so
that in the middle of the 60's inflation almost reached the
hyperinflationary level. The element that contributed to the slow
accommodation of the real government revenues to the price increase
was the inefficient method of tax collection, the structure of the tax
revenues where the indirect taxes and taxes of international trade
prevailed. The econometric analysis showed that for the
accommodation of the tax level to the price level it took 4.5 months on
the average, whereby the expenditures reacted to the price increase

84
with an average fag of only 1.5 months, that is three times faster.
Therefore, the Indonesian case showed that when the prices increase,
the real taxes follow them harder and harder, which results in
increasing deficit.

Argentinean experience

The fall of real tax revenues in Argentina, in the period 1968 - 1976
was analysed by Tanzi (1978). His analysis is limited solely to the
revenue side of the budget, but it takes into account "normal" tax
revenues as well as the inflation tax revenues, and deals with the
behaviour of their sum. From the beginning until the end of the period
considered, inflation climbed from one-digit level to the three-digit
level, that is to 400% percent. The price development reflected in the
development of the real tax revenues, which, in that period of time fell
from almost 20% share in the GDP to 12-13% of GDP in the period of
the highest inflation.

Tanzi described the relationship between the price increase and the fall
of the real tax revenues with the expression according to which the
total revenues are TRB = Bm0e-"B + (T0/(1+B)n). He simulated the
development of the total revenues by taking the initial share of taxes in
GDP, which amounted to 20% or 0.2, as the tax burden at zero
inflation (T0). For the coefficient m0 (the share of the money balances
in the GDP) he used its preinflationary level of 15%, that is 0.15. He
calculated that the weighted average collection lag for the whole period
was 5 months. Tanzi estimated econometrically the value of the
parameter " (the sensitivity of the demand for money with respect to
the rate of inflation), and found out that its value changed in that
period, that is, in the period with low inflation it was 2.458, and in the
years with high inflation it was 0.564. By introducing the values of the
needed parameters in the formula for total revenues, Tanzi showed
that the actual total real revenues of the government followed almost
the same pattern as his formula predicted. The "normal" tax revenues
were falling along with the increase in inflation, while the inflation tax

85
revenues were increasing in the first period, and decreasing later, as
the consequence of the fall of the demand for the real money balances.
At the same time Tanzi explained that the maximal attainable real
government revenue depends on the coefficient of the sensitivity of the
demand for money with respect to the rate of inflation. At low rates of
inflation, the demand for money reacts heavily to the fall in the real
value of money, so it showed that in the beginning of the development
of the inflationary spiral, due to the narrowing of the base of the
inflation tax, the total revenue-maximizing rate of inflation was 15%.
Later, at the high level of inflation the ratio of money to GDP becomes
so low that it was impossible for it to get any lower, and the sensitivity
of the demand for money with respect to inflation decreased. Under
such circumstances the total revenue maximizing rate in Argentina was
100%. At rates higher than 100% the erosion of the collection of the
real "normal" revenues would be so strong that it would be producing
negative effects that offset the positive effects of the inflation tax
revenues.

The Bolivian experience

Very strong Olivera-Tanzi effect was noticed in the Bolivian

hyperinflation in 1985-86 (Sachs). After the World War II it was the
only one hyperinflation in the sense of the Cagan definition of
hyperinflation, and one of the highest inflations ever. In the first half of
1985 inflation amounted to 26000%, and from May to August 1985
even 60000%. The cause of that very high inflation was the complete
loss of the credibility of the government incurring debt crisis, and it was
directly initiated by the imbalance in the balance of payments, caused
by the price decrease of the Bolivian export products in the
international market. Before the inflation started, the Bolivian tax
system was very weak. The former governments accumulated deficits
and the problem of how to finance them had been solved by foreign
indebtedness. When it was not possible any more, governments shifted
to money creation. As the time passed by, inflation got an incredible
acceleration, while, at the same time, the real inflation tax revenues

86
failed to increase on one hand, and the real collection of all other taxes
collapsed. In spite of the sharp decrease in the government
expenditures, the deficit became the persistent problem. With the
beginning of implementation of the stabilization programme at the end
of 1985 the real tax revenues of the central government rose
considerably. The main reason was the effect of the devaluation of
pesos on the prices of the energents and, according to that, the
increase of the ability of public enterprises (state oil company) to pay
taxes. Already in the following year the budget was balanced, and it
even became slightly positive.

The Bolivian experience is just another example of how easy it is to get

into the whirlpool of inflation and fiscal deficit, that is, how easy the
Olivera-Tanzi effect starts to operate, but they show as well that the
increase in real government revenues usually occurs once the price
stabilisation has been started.

It can be observed that the relationship between the real tax revenues
and the rate of inflation has been noticed, but there was different
attention given to it, or it was analysed differently. The conclusions
about that relationship have been reached upon the insight into the
data on the development of the rate of interest and the real tax
revenues, on the basis of the tests of simple regressions, but also by
constructing more complex models with simultaneous equations.

The results of the analysis, regardless of the analytical method, show

that in the conditions that lead into inflation, the interrelations between
the level of real tax revenues and the rate of inflation are being
induced. The examples imply that the fiscal deficit, no matter where it
does stem from, is in the majority of cases the initiator of those
unfavourable developments. The deficit is, thus, the necessary, but not
the sufficient condition for the Olivera-Tanzi effect to happen. The
sufficient condition is that there is no possibility for the government to
cover the deficit in noninflationary ways. The negative relationship
between the real tax revenues and inflation is the stronger the more

87
regressive the tax system as a whole, that is, the more is it based on
the indirect and specific taxes, the weaker the tax administration, the
looser the fiscal discipline of the tax payers, and the more difficult the
tax system accommodates to the price increase (by indexing the tax
bases, by increasing the tax rates and by shortening the collection
lags).

As for the indexation of the tax bases, it has to be emphasized that

some countries (Israel for example), remained uninfected by the
domino-effect of the inflation and deficits, thanks to the introduction of
the system of comprehensive indexation at the right time (Sadka,
1990). But, such a way of fighting the Olivera-Tanzi effect succeeds
only by chance and in relatively stable economies. Therefore, there is a
general recommendation to keep away from taking inflation as remedy
for satisfying the increasing needs of the government, or, what seems
even better, to keep the government consumption on those levels that
the government can afford.6

6
The aforementioned examples could be misused, or they could be misunderstood.
They are all cases of very high inflations and hyperinflations, so they could lead to a
conclusion that there is a danger only if there is an excessive inflation. On the
contrary, one has to infer differently. The cases of very high inflations and
hyperinflations are being considered because hyperinflation, like the picture under
the microscope, allows us to see some things clearer then it could be possible in
some less extreme cases. What can be seen in the moments of hyperinflation can
easily happen even when the intensity of the price increase is not so strong. So one
has to take the advice given by Choudhry (1993) which says that, thanks to the
causal relationship between the deficit and inflation, the sustainable level of deficit
is much lower than it has usually been thought. In the same way, even in small
doses, inflation is a very dangerous way of collecting taxes, because very low
inflation can also have negative influence on the budget balance, the higher the tax
burden in the country (the share of the taxes in GDP) and the lower the ratio of
money to GDP.

88
4. OLIVERA-TANZI EFFECT IN THE CROATIAN
STABILIZATION PROGRAMME OF 1993

The theoretical framework given by Julio G. Olivera and Vito Tanzi on

the effect of inflation on inflation tax revenue, real "normal" and real
total taxes gives clear determinants of the fiscal policy in the period of
the implementation of the stabilization programme. The classical
Olivera-Tanzi relationship between inflation and real tax burden (Figure
2) is the one when the maximum of the share of the inflation tax in the
GDP (IR) is reached at the rate of inflation BA (point A), the ratio of the
real normal taxes in GDP (T) falls together with the inflation, and
maximum of the real total taxes (TR) is therefore reached at lower rate
of inflation, BB (point B). This shape of "normal" tax curve is
characterised by the tax system without indexation, in which the
collection lags cause the inflationary devaluation of the real tax
revenues, which, at unchanged level of the real expenditures results in
fiscal deficit. The deficit is the smaller the lower the inflation rate, and it
is predominantly financed by inflation tax. In the interval between BA
and BB the decrease of deficit caused by disinflation is greater then the
loss of the inflation tax. The same levels of the real fiscal revenues
(points C and D, for example) can be attained at the different rates of
inflation, BC and BD, respectively. Although lowering of the rate of
inflation from BC to BD would be neutral from the fiscal point of view,
those two combinations have completely different macroeconomic
implications. At point C real expenditures are covered by normal taxes
and by inflation tax, that means there is some deficit, while at the point
D the deficit is being eliminated, as well as the inflation tax as the
source of its financing.

The fiscal position, such as in the point D, is the policy known as

"financing out of real sources". Since the inflation tax is equally "real"
as any other tax (Dornbusch, 1991), such a mixture of financing should
probably be named "financing out of stable sources". The emphasis in
the analysis of the Olivera-Tanzi effect should be on the stability of the
sources of finance.

89
The effect of disinflation on fiscal revenues of a country depends on the
factors that determine the position, the shape and the structure of the
curve of the total real tax revenues, and that are the characteristics of
the fiscal, monetary and the whole economic system of a country.

In the countries passing the phase of disinflation the points of

maximum of inflation tax and total taxes change. That means that such
processes cause some shifts in the possible sources of financing
government expenditures, and that the way how the monetary
phenomena affect fiscal phenomena change, as well. The Tanzi's
findings for Argentina, that the total fiscal revenue-maximizing rate of
inflation was considerably lower at the lower than at the higher inflation
rates, show clearly that there was some shift of one or both
components of the real total taxes. Structural shift of the curves in the
two-dimensional space determined by the inflation and the fiscal
revenues could be attributed to some third variable. In the inflation tax
component that could only be the variables that are not included in the
Tanzi's specification of the demand for the real money balances.

The empirical analyses of the Croatian stabilization show that under the
strong currency substitution the exchange rate is one variable causing
structural shift of the inflation tax curve. If the country implements a
heterodox antiinflationary programme, the intensity of the reverse
currency substitution is the stronger, the higher the credibility of the
programme. The shifts of the normal taxes can be influenced by the
reverse currency substitution as well, but also by those essential
changes in the fiscal system that usually follow the heterodox
antiinflationary programmes.

90
The money demand function applied by Olivera and Tanzi, as well as by
the others afterwards could be characterised as a long term money
demand function, based on very rigid assumptions of unitary income
elasticity and of the zero elasticity of demand for money with respect to
the other alternative measures of the opportunity cost, like exchange
rate for example.

In analysing the demand for money at high inflation rate the short time
dynamics of the accommodation of the actual level of money balances
to the preferred level ofmoney holdings is of great importance (Hwang,
1985). The sensitivity of the demand for money in short term depends
also on the velocity of that accommodation. However, the most serious
shortcoming of the specification of the money demand function used is
the omittance of the opportunity cost of holding money. The theoretical
analysis shows that in the full specification (including the elements of

91
the opportunity costs like interest rate and/or exchange rate) of the
demand for money in high inflation and high currency substitution
countries, the coefficient at the rate of inflation would "hide" the
parameters of the accommodation of the demand for money and the
parameters of "money illusion" (Anušić, Rohatinski and Šonje, ed.,
1995).7 Similarly, the coefficient at the opportunity cost of the
alternative financial instruments would describe the additional elements
like the strength of the effect of substitution of the domestic with the
foreign currency and/or with the interest holding deposits. If the
opportunity cost is tightly linked to inflation (or indexed), like this is
often the case in the high inflation countries, then the total influence of
inflation on demand for money is reflected in several parameters, not
only in the coefficient ". In this way, all those effects are implicitly
included in the coefficient ". By neglecting opportunity costs from the
specification of the money demand function, the estimate at the
inflation rate reflects and "hides" a part of the interest rate and/or
exchange rate elasticity of the demand for money, too.

The econometric estimation of the full specification of the demand for

money in the Republic of Croatia showed that in the period after the
stabilisation measures have been introduced, a significant increase in
the elasticity of the demand for money with respect to the interest rate
and with respect to the change of the exchange rate happened (Anušić,
Rohatinski and Šonje, ed., 1995). In other words, the fall of the
nominal exchange rate in the period of stabilisation is much more
reflected in the increase in money demand then it would have been the
case in the period before the stabilisation programme. Equally, if there
was the depreciation of exchange rate in the prestabilisationary
manner, the decrease of the demand for the domestic currency would
be four times greater than in the period before stabilisation.

7
The parameter of "money illusion" descrihes the sensitivity of the preferred level
of money with respect to inflation (Patinkin, 1965: Gapinski, 1982).

92
The econometric estimation of the inflation tax in Croatia in the period
before and after stabilisation.

In this paper the specification of the money demand function in Croatia

is based on the theoretical background of the short term money
demand function in the presence of currency substitution and without a
priori estimation of the coefficient of "money illusion" and the pattern of
the accommodation of the actual level of money holdings to the
preferred level. The specification includes the income variable without
the a priori assumption on the unitary income elasticity of money
demand. This hypothesis will be tested econometrically. Furthermore, in
contrast to the theory, the specification does not include the elements
of the opportunity costs (like interest rate or the rate of change of the
exchange rate). It is expected that the omission of those variables
would result in higher price elasticity than the one in the full
specification. The used form also contains the variable of the structural
shift of the inflation rate in the period after stabilisation in order to test
the hypothesis that the revenue-maximizing rate of inflation changed
after the stabilisation programme has been introduced. The
econometric estimation of such money demand function in the period
January 1991 - October 1994 gives us the following results:

Imt = 0.0443 + 0.8505 Imt-1 + 0.1587 IYt - 0.5698 INFLAt

(0.55) (30.47) (2.52) (-6.72)
- 2.6270 (INFLAt*DUM9311Xt) - 0.0303 DUM1t - 0.0424 DUM2t
(-1.31) (-0.79) (-1.12)
+ 0.0027 DUM4t - 0.0210 DUM5t - 0.0052 DUM6t + 0.0696DUM7t
(0.07) (-0.58) (-0.14) (1.87)
+ 0.0149 DUM8 t - 0.0320 DUM9 t + 0.0389 DUM10 t
(0.41) (-0.87) (1.05)
- 0.0054 DUM11 t + 0.0582 DUM12 t
(-0.13) (1.49)

RBAR2 = 0.9918; DW = 1.85; SEE = 0.0507; F = 365; Period: 91:1-94:10

93
where:

Im = the logarithm of the real value of M1 at the end of month,

Im t = In (M1 t/P t), where P, stands for the implicit deflator of GDP,
IY = the logarithm of the real GDP,
INFLA = the rate of the change in the implicit deflator of GDP,
INFLA t = (P t/P t-1 -1),
DUM9311X t = binary variable for the period 1993:11-1994:10;
DUM9311X = 1 in that period, and DUM9311X = 0 in all other months,
DUM x = binary variable for the month X; DUM x = 1 in the month X,
and DUM x = 0 in all other months.

The estimated elasticity of the demand for money in the period before
stabilisation was -0.5698*B and for the period after stabilisation -
3.1968*B. This means that the revenue maximizing rate of inflation in
the period before stabilisation is:

B*pre = 175.5% a month,

and in the period after stabilisation:

B* post = 31.3% a month.

The estimated income elasticity of the money demand is 1.06, that is

close to the upper boundary of the elasticity in the Baumol-Tobin
theory of the money demand, as well as to the assumption taken by
Tanzi in his theoretical framework.8

The shift of the inflation tax curve in the period after the stabilisation
can be seen on the Figure 3.9

8
By dividing the estimated equations by the income level the original form of the
equation used by Tanzi in his paper could be obtained.
9
In this paper we give the graphics of the Olivera-Tanzi effect with the rate of
inflation on the X axis and with the tax burden on the Y axis.

94
The values of both curves are obtained on the basis of the estimated
functions of money demand. In the period before stabilisation, inflation
over 30% which prevailed at that time, would result in additional real
inflation tax revenues. According to the estimated equation, the level of
revenue maximizing inflation in Croatia stood at 175.5% monthly.
Under such circumstances it would, however, be uncertain whether the
demand for money would have fallen together with inflation (the
elasticity was -0.5698*B) or would that mechanism have started to
accelerate. Then, the B*pre would be lower than the estimated one.

In the period after the stabilisation there was a considerable change in

the possibility to finance government expenditures with inflation tax.
The maximum of the inflation tax revenues can now be achieved at
much lower inflation rate (31.3% monthly) and the decrease of the
maximum level from 67% of the GDP in the period before stabilisation
to the 12% of GDP in the period after stabilisation leads to a conclusion

95
that there is not much room for inflationing the Croatian economy. To a
stronger inflationary shock the Croatian economy would react with
heavy substitution of domestic currency. In this way, the reversal of
inflation to its stabilisation level would result in lower inflation tax, and
at higher rates of inflation, even in its absolute decline.

Econometric estimation of the normal tax in Croatia in the period before

and after stabilisation.

The normal tax function follows the Tanzi's assumption of variable price
elasticity of the real revenues from "normal" sources, with an additional
structural shift variable in the period after stabilisation. The equation is
estimated for the period December 1991 - December 1994:

In (Tt/Yt) = -1.0178 - 0.7759 INFLAt - 2.1870 (INFLAt*DUM9311Xt)

(-28.70) (-4.62) (-0.61)

RBAR2=0.4256; DW=1.71; SEE=0.1254; F=74; Period: 91:12-94:12

where Tt/Yt denotes the ratio of tax revenues in the Croatian GDP
accounted on the cash basis, and INFLAt and DUM9311Xt denotes the
rate of inflation and the structural shift variable, respectively, like in the
last equation.

The estimates show that at zero rate of inflation, there is the real tax
burden of the GDP that amounts to 36%. Any price increase causes the
erosion of the real tax revenues lowering this ratio. This statement
holds in the period before stabilisation, and for the period after the
stabilisation as well. In the period before stabilisation the price elasticity
of real tax burden of the Croatian GDP was -0.7759*B. In the period
after stabilisation the estimated price elasticity is -2.9629*B , although
with unsatisfying statistical properties.

Figure 4 illustrates the estimated values of the share of the normal

taxes in GDP in the period before and after stabilisation. Steeper slope

96
of the curve in the period after stabilisation is the consequence of
higher price elasticity in that period. These findings imply the possibility
of the functioning of the "ratchet effect" in the relationship between the
inflation and the tax revenues. Lowering inflation will result in the
increase of the real tax revenues and in lower budget deficit, while a
subsequent increase in inflation to the initial level would cause even
higher deficit then the one which could have been possible in the initial
period. The moral of this result for the fiscal policy in Croatia is that any
renewal of the inflation at the prestabilisationary level of about 30% a
month would be disastrous for the public finances, and for the
economic system as well.

Figures 5 and 6 illustrate the values of the total real tax burden and its
components in the period before and after the implementation of the
stabilisation programme. The rate of inflation maximizing total potential
fiscal burden in the period before stabilisation was 140.2% a month

97
(Figure 5). At this rate of inflation there would be a great loss of the
real tax revenues, whereby the ratio of the real tax revenues in the
GDP would fall to only 11.3%. This could result in huge deficit financed
wholly from an unstable and uncertain source as the inflation tax.

However, in the period after stabilisation, the situation is completely

opposite. The maximum of the fiscal burden can now be achieved at
the zero rate of inflation. Although the inflation tax maximizing rate of
inflation now amounts 31.3% monthly, as we already emphasised, the
inflationary loss of the normal taxes exceeds the inflation tax gains.

98
99
5. CONCLUSION

Inflation was not to the benefit of the Croatian fiscal policy neither
before the implementation of the stabilisation programme, nor in the
poststabilisation period. The analysis showed that in Croatia, inflation
always has negative effects on the revenues from the normal taxes,
that is, every inflation opens big possibilities for entering into the fiscal
deficit.

For Croatia, the better economic future will certainly come. However,
the road to that future will be to a great part determined by the
economic policy of today. The empirical finding of this paper is that the
inflation tax in the period after stabilisation in Croatia could be
maximized at high rates of 31.3% monthly. For some, it can serve as
an argument in favour of a statement that inflation is needed, but this
is just one side of the medal. Inflationary erosion of the normal taxes is
much stronger then the inflationary component of total taxes. Any
involvement in inflationing of the Croatian economy in order to
"finance" faster economic activity would result not only in a loss of the
real tax revenues and in deficit, but it would also have the opposite
effect - the loss of the total real fiscal revenues. Price stability
guarantees the stability of financial sources for the economic recovery
that is about to start. There is simply no other choice for the Croatian
fiscal policy.

100
BIBLIOGRAPHY

Aghevli, Bijan B. and Mohsin Khan (1977): Inflationary Finance and the Dynamics of
Inflation: Indonesia, 1951-1972, American Economic Review, Vol. 67, pp. 390-403.

Aghevli, Bijan B. and Mohsin Khan (1978): Government Deficits and the Inflationary
Process in Developing Countries, IMF Staff Papers, Vol. 26, pp. 383-416.

Anušić, Zoran, Željko Rohatinski and Velimir Šonje (ed.)(1995): A Road to Low
Inflation: Croatia 1993-1994, The Government of the Republic of Croatia, Zagreb.

Cagan, Phillip (1956): The Monetary Dinamics of Hyperinflation, In The Theory of

Inflation, Michael Parkin (ed.), pp. 185-278.

Canavese, Alfredo J. and Daniel Heymann (1992): Fiscal Lags and the High Inflation
Trap, Quarterly Review of Economics and Finance, Vol. 32, No.2, pp. 100-109.

Choudhry, Nurun N. (1990): Fiscal Revenue and Inflationary Finance, IMF

Working Paper, WP/90/48.

Dornbusch, R. (1991): Experiences with Extreme Monetary Instability, in:

Commander, S., ed., Managing Inflation in Socialist Economies in Transition., EDI
Seminar Series, Washington, D.C.: The World Bank: 175196.

Franco, Gustavo H.B. (1990): Fiscal Reforms and Stabilization: Four Hyperinflation
Cases Examined, The Economic Journal, 100, pp. 176-187.

Gapinski, J.H. (1982): Macroeconomic Theory: Statics, Dynamics, and Policy,

Economics Handbook Series, New York: McGraw-Hill.

Hwang, H. (1985): Test of the Adjustment Process and Linear Homogeneity in a

Stock Adjustment Model of Money Demand, Review of Economics and Statistics, 67:
689-692.
Kiguel, Miguel A. (1989): Budget Deficits, Stability and Monetary Dynamics of
Hyperinflation, Journal of Money, Credit and Banking, Vol. 2, No. 2, pp. 148-
157.

Nestić, Danijel and Velimir Šonje (1994): Stopping High Inflation in an Exsocialist
Country: the Case of Croatia 1993/1994, Croatian Economic Survey, pp. 19-61.

101
 Olivera, Julio H.G. (1967): Money, Prices and Fiscal Lags: A Note on the Dynamics
of Inflation, Banca Nazionale del Lavoro Quarterly Review, No. 82, pp. 258-
268.

Patinkin, D. (1965): Money, Interest, and Prices., New York: Harper and Row.

Sachs, Jeffrey (1986): The Bolivian Hyperinflation and Stabilization, Working

Paper No. 2073, NBER, Inc., pp. 1-47.

Sachs, Jeffrey and P. Larrain (1993): Macroeconomics in the Global Economy,

Harvester, Wheatsheaf.

Sadka, Efraim (1990): An Inflation-proof Tax System? Some Lessons from Israel,
IMF Working Paper, WP/90/46, pp. 1-23.

Tanzi, Vito (1978): Inflation, Real Tax Revenue, and the Case for Inflationary
Finance: Theory with an Application to Argentina, IMF Staff Papers, Vol. 25, No.
3, pp. 417-451.

Végh, Carlos A. (1992): Stopping High Inflation, An Analyticial Overview, IMF Staff
Papers, Vol. 39, No. 3, pp. 626-695.

102

View publication stats