Shortage and Currency Substitution in Transition Economies:
Bulgaria, Hungary, Poland, and Romania

International Advances in Economic Research 7:2 (May 2001)

Robert F. Mulligan
Western Carolina University and the State University of New York at Binghamton
Department of Economics, Finance, & International Business
College of Business
Western Carolina University
Cullowhee NC 28723 U.S.A


Erwin Nijsse
Royal Dutch Shell


 This paper examines currency substitution in four central European countries, Bulgaria, Hungary, Poland, and Romania, during the end of central planning and transition to market economies.  Before liberalization, central European economies faced increasing shortage and repressed inflation in the official sector, and growing free, grey, and black-market sectors.  The population held substantial portions of their wealth in real assets and foreign currency.  Furthermore, part of savings was held as hunting money against potential opportunities to buy large quantities in official stores, or pay price premia on the black market. The shift from centrally-planned to market economies is modeled by including a shortage variable.  Foreign currency demand and consumption functions are estimated by the Johansen procedure.  Official-sector shortage is found to be an important determinant of foreign currency demand in each country.

1.  Introduction
 This paper examines currency substitution in four central European economies, Bulgaria, Hungary, Poland, and Romania, from 1986 to 1994.  Currency substitution is an important concept in the history of economic thought.  Menger (1874, pp. 268-270) cites the instance of parallel exchange commodities used in ancient Mexico during the transition from a pure barter to a monetary economy.  Interestingly, currency substitution can therefore predate the establishment of money.  The phenomenon of currency substitution is a general precondition of which Gresham's law is a special case:  at least two currencies or exchange commodities must circulate in parallel before one ("bad money") can drive out the other ("good money.")   Gresham's law can thus be understood as an extreme case of currency substitution resulting ultimately in abandonment of one of the currencies.  In Gresham's law, the good money's superior store-of-value character leads to hoarding and the bad money's equal value for transactions leads to it being used for all transactions.
 In centrally-planned economies (CPEs), currency substitution was a very singular phenomenon.  Gresham's law never became fully operable because although foreign currency (the "good money") had a superior store-of-value character, both foreign and domestic currencies had separate, complementary transactions uses in official and non-official sectors. In most eastern European economies, the rate of currency substitution seems strongly correlated with high inflation and uncertainty.
 Experience of the transition economies closely resembles the recent history of some Latin American countries, like Argentina, Bolivia, Mexico, Peru and Uruguay (Sahay and Végh, 1995).  Currency substitution in Latin America received a great deal of attention in the literature (see e.g. Calvo and Végh (1992), Rodriguez (1992), Rojas-Suarez (1992), Savastano (1992), and Sturzenegger (1992)).  Until recently, currency substitution in the transition economies of central and eastern Europe received much less attention, with the notable exceptions of Charemza and Ghatak (1990), Aarle (1994), and Aarle and Budina (1995).
 The remainder of this paper is organized as follows. Section 2 discusses the economic theory of currency substitution.  Section 3 analyzes data for the four countries and presents shortage measures.  Section 4 presents empirical estimates of the foreign money demand function derived in Section 2.  Section 5 summarizes main points and draws conclusions.

2.  Background and Literature
 The traditional view of CPEs as "shortage economies" (Kornai, 1980), states that central planning authorities arbitrarily divide the money stock into household and state-enterprise sectors (Hartwig, 1983, 1985).  In each sector, money flows play only a passive accounting role, enforcing adherence to the central plan.  Money demand does not result from utility-maximizing agents' choice, but as a by-product of central planning.  Central planners typically chose a non-market-clearing price vector, leading to excess demand and forced savings.
 This traditional view, however, applies only to highly-idealized CPEs without private sectors, or free, grey, or black markets to equilibrate the economy.  Unlike other economic agents in CPEs, households faced hard budget constraints and always tried to maximize utility from real and financial assets.  Consequently, savings of the population in CPEs were voluntary, never forced.  It is logical to view their choice of money holdings as resulting from constrained-optimization.
 The range of alternative financial assets was broader in reality than the traditional view assumed.  Although households did not hold either foreign or domestic bonds, real asset and foreign currency holdings were often substantial.  Foreign currency holdings consisted of private hoards (mattress money,) deposits in banks abroad, and where permitted, (as in Poland and Yugoslavia,) domestic deposits of foreign currency.
 These general observations about household behavior in CPEs suggest the importance of environmental constraints.  Shortage in the official sector, brought about by the central planning authorities' arbitrary selection of a non-market-clearing price vector, is the most important feature of these economies.  In addition, availability of goods in the black market, and the possibility of holding foreign currencies and real assets as stores of value, must also be acknowledged.  Households should be viewed as utility-maximizing agents choosing optimal money holdings subject to constraints.  This view contrasts strongly with the traditional one where households passively adapt their spending to goods availability.
 Currency substitution is examined using quarterly data for Bulgaria, Hungary, Poland, and Romania.  Due to constraints on data availability the study is not extended to more countries.  Although many countries, including Russia, now record data on foreign currency holdings, the time span of data available before the end of central planning is too short for econometric estimation.
 Mulligan and Nijsse (1995) present a model where an official-sector cash-in-advance constraint is augmented with a shortage variable.  The model includes transactions, precautionary, and portfolio motives for holding money in a centrally-planned economy in transition.  The transactions motive is incorporated through two cash-in-advance constraints: one for the official sector, where only domestic currency can be spent; and one for total currency holdings similar to Lane (1992), and  Nijsse and Sterken (1995) who modeled the demand for money in Poland.
 The precautionary motive is incorporated through shopping time, which is a function of purchases in each sector, real balances of the two currencies, and shortage.  Shopping time eats away at leisure and has a negative impact on utility.  This captures the behavior Kornai (1980, 457-458) described as purchaser alertness.  Facing shortage, consumers carry shopping money to take advantage of unpredictable, and infrequent, opportunities to purchase large quantities at low prices in the official sector.  Both currencies are also stores of value, and holdings also depend on their rates of return relative to goods.
 If the shortage-augmented official-sector cash-in-advance constraint, which binds as an equality, is used to marginalize with respect to domestic money, the demand for foreign money is expressed as a function of the black market exchange rate, aggregate household consumption, inflation, and shortage in the official sector.  The functional form of the foreign-money demand equation depends on the form of the utility function, which is unknown and unobserveable:

(2.1)   mf  = mf (c, p, s, x)
 This foreign currency demand function is the basis for empirical estimation in section 4 below. The empirical method is the Johansen maximum-likelihood procedure, which is used to test for cointegrating relationships and estimate the cointegrating vectors.  The Johansen likelihood ratio test is used to test the shortage variable for significance and restrictions which remove it from the system; shortage is always found to be significant.

3.  Data
 Contrary to what is generally assumed, the four economies in this study practised only partial, not full, central planning.  Reforms had already been started by the 1970s, especially in Hungary and Poland (Jeffries, 1993).  In Hungary, the "New Economic Mechanism," aiming at greater enterprise autonomy, partial price liberalization, and flexible wages, was implemented as early as 1968.  In Poland, reforms in the 1970s were modest, but accelerated in the early 1980s.  In Bulgaria and Romania, credible reforms were implemented only in the second half of the 1980s, and until late 1990, both economies remained highly centralized.
 Following the reform programmes, the private sector in the four CPEs expanded.  During the 1970s private sector activities were dominated by agriculture.  In the 1980s the private sector broadened in scope to include small retail stores, hotels, restaurants, and handcrafts.  Next to the legalized private sectors, black markets developed in several consumer goods, including durables like flats and cars, as well as foreign currency and precious metals.  These markets not only satisfied excess demand from the official sector, but had a speculative character.  Real assets were good investment alternatives to bank deposits in CPEs.  Especially in Poland, black-market activity was widespread and closely followed by the authorities (Wyczanski, 1989).  The existence of legal and illegal private sectors had a positive impact on these economies, playing an equilibrating role by allowing households to adjust spending and savings to desired levels.  However, at the same time, the relative profitability of private sector activities induced reallocation of resources from the socialized sector, preventing the socialized sector from functioning properly.
 The financial sectors were liberalized in each of the four countries at the end of the 1980s, before overall liberalization.  This happened in Hungary in 1987, in Poland in 1989, and in Bulgaria and Romania in 1990 (Calvo and Kumar, 1993).  Traditional Monobank systems were reformed into western-type two-tier banking systems.  Interest rates on bank deposits, which were traditionally fixed and often negative in real terms, were gradually liberalized.  With the implementation of monetary stabilization programmes, positive real interest rates became an important instrument to support tight monetary policies.  At the same time, household savings in the form of financial assets greatly expanded.  In Hungary, for example, the household savings rate increased from the traditional pre-reform level of 2-4%, to 14% in 1992 (Bod, 1994).  The strong increase in savings rates was probably due more to the changing economic situation, than interest rate policies of national banks at the end of the 1980s.  Before liberalization, people felt secure in terms of employment security and pensions, but perspectives changed dramatically as the CPEs entered transition.  The transition economies still face political and economic instability, high inflation, high unemployment, and only partially-indexed social security allocations.

Foreign currency holdings
 While the macroeconomic environment affects demand for foreign currency, the supply of foreign currency depends critically on institutional constraints.  Poland and Yugoslavia, with fairly liberal banking systems, saw higher dollarization rates before liberalization than after.  In other countries, the absence of significant dollarization was primarily due to tight controls on foreign exchange and the banking system.  Foreign currencies were mainly supplied by smuggling and were held almost exclusively as "mattress money," which cannot be estimated reliably.
 In Bulgaria at the end of the 1980s, foreign currency as a percentage of total money holdings (M2), increased from approximately 5% in 1986, to above 34 percent by the first quarter of 1994.  Hungary, having much lower inflation, saw far less currency substitution.  However, the percentage of foreign currency deposits, 20% of M2 at the beginning of 1994, shows an increasing trend.
 In Poland, currency substitution was a much more significant phenomenon, and the presence of an active black market suggests Poles held foreign currency for transactions reasons.  From 1985, foreign deposits as a percentage of M2 increased strongly, reaching its record high at the end of 1989, 75% of M2.  In contrast to the other countries, Poland saw "resubstitution" of foreign currencies back into zlotys at the end of the 1980s, due to falling inflation and the rigorously-followed policy of positive real interest rates on zloty deposits.  Figures may exceed official numbers because of the use of the black market exchange rate, but considering the near-equality of the official and black market exchange rate from March 1989, the general pattern is not dramatically different.
 However, among the four countries considered, Poland was a special case because from the beginning of the 1980s, zloty inflation was high and foreign currency deposits could be freely held without a declaration of source (Chawluk and Cross, 1993).  Thus, in Poland, currency substitution was already extensive before the end of central planning.  In the other countries, currency substitution faced greater restrictions.
 In Romania, finally, foreign currency holdings displayed a strongly increasing positive trend following liberalization of the banking system at the end of 1990, representing approximately 35% of M2 in the beginning of 1994.  Comparing experiences of Romania, Bulgaria and Hungary, with Poland, suggests restrictions on holding foreign currency deposits in the former three countries were relatively severe under central planning.  There appears to be a generally strong correlation between macroeconomic instability, represented by inflation, and foreign currency holdings.  Bulgaria and Romania, with annual inflation rates of 73% and 256% in 1994 (EBRD, 1994), show strongly rising foreign currency deposits.  In Poland and Hungary, the situation seems to have stabilized, although in 1994 the share of foreign currency holdings in both countries was slightly higher than in 1993.
 As Aarle and Budina (1995) note, currency substitution or "dollarization" is likely to remain a significant phenomenon, even if the economies in transition succeed in achieving macroeconomic stabilization.  Empirical evidence from Latin America supports this conclusion (Savastano, 1992).  For the four Central European economies, foreign currency as a percentage of M2 appears to have stabilized only for Poland.  Romania and, to a lesser extent, Bulgaria and Hungary, show increasing positive trends.

 The analysis of shortage in CPEs has been thoroughly analyzed (notably by Portes and Winter (1980), Portes (1981), Charemza and Quandt (1982), Jansen (1982), and Kornai (1980).  In these papers excess demand is either directly or indirectly measured.  In practice this leads to highly different results, (for example, for Poland see Charemza and Gronicki (1988) and Podkaminer (1982, 1989)) and simpler measures perform fairly well (Charemza, 1992).  Various shortage indicators have been used in the literature: Hartwig (1987) uses the reciprocal of the velocity of circulation; Payne (1990) uses real consumer credit; Feltenstein et al (1990) use the part of savings explained by virtual prices, a combination of official prices and real money; Lane (1992) criticizes earlier measures as circular and uses instrumental variables to model expectations.
 Nijsse and Sterken (1995) use four alternative measures for Poland: first, the ratio of nominal household income growth to the growth of nominal net material product corrected for imports and exports; second, the price differential of the free market price index over the official food price index; third, the ratio of nominal income to the growth of official retail sales (after Chawluk and Cross, 1993); and fourth, the ratio of consumption in the official sector to inventories held in this sector.
 Data constraints forced the choice of a simple but highly plausible shortage indicator which has been shown to work well empirically: the ratio of the black-market over the official exchange rate (Peebles, 1991).  As with any shortage indicator, this one presents potential problems.  For the period before liberalization, the black-market exchange-rate was overvalued in most CPEs, whereas the official exchange rate was undervalued and followed developments in the black-market exchange rates slowly, with a long and variable time lag.  Shortage rose in all four countries in the latter half of the 1980s, and in the case of Bulgaria and Hungary, fell somewhat at the beginning of 1990, before liberalization.
 At the moment of price liberalization the shortage indicator is set arbitrarily at 1, since following price liberalization the Walrasian auctioneer is assumed to work again, and in any case, the central planning authorities stop fulfilling their function of selecting the non-market-clearing price vector.  The dates of liberalization were taken from Aarle and Budina (1995): for Bulgaria, January 1991; Hungary, February 1991; Poland, January 1990; and Romania, April 1991.
 Shortage in Hungary was low compared to the other countries, with Bulgaria having the worst shortage, in the third quarter of 1989.  This confirms earlier evidence found by Charemza and Ghatak (1990), for example.  In Hungary, prices followed market developments, except for a basket of basic consumer goods that remained subsidized until the end of the 1980s (Jeffries, 1993).
  For Poland, shortage figures suggest a positive correlation between foreign currency holdings and shortage; for the other three countries the correlation seems negative.  This is due in part to differences in banking regulation under central planning, specifically that foreign currency deposits were allowed in Poland, as discussed above.  In addition, during partial central planning in Poland, the transactions value of foreign currency on the black market was much larger than in other countries, where foreign currency seemed to be held mainly as a store of value.  The evidence found by Charemza and Ghatak (1990) for Hungary and Poland supports this conclusion.  Bulgaria, Poland, and Romania all seem to suffer more severe shortage.  Evidence suggests the scope for black market trade was by far largest in Poland.

Exchange rates
 All four countries in this study show steady depreciation of the domestic currency.  By 1994, all countries had implemented internal current-account convertibility, except for tourism, where restrictions remained on amounts people were allowed to take abroad.  In general, capital-account convertibility has not been implemented, but restrictions on profit repatriation have loosened.
 In Bulgaria, Hungary, and Romania, a spread between black market and official exchange rates could be observed as late as the beginning of 1994.  This indicates restrictions remained on trade in foreign currencies, or that central banks intentionally created a spread.  In Romania, for example, the interbank foreign exchange rate was unified with the black market exchange rate only in 1994.  The Romanian leu has floated freely ever since.  In Bulgaria, the black market exchange rate was quoted as 73 lev per USD versus the official rate of 65 lev per USD in the first quarter of 1994.  The lev has been floating freely since February 1991 and is internally convertible.

 By 1994, the domestic currency in Hungary, the forint, was pegged to a basket of currencies consisting of 70% ECU and 30% USD.  In Poland, the zloty is pegged to a basket of currencies as well, and depreciates at a pre-announced rate of 1.5% per month.  Trade in foreign currencies was fully legalized as early as the beginning of 1989 and the spread between black-market and official exchange rates nearly vanished after the devaluation of the zloty on January 1, 1990.

4. Empirical Results

Unit Root Tests
 Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests were applied to check whether the single variables contain one or more unit roots.  Both the Lagrange-Multiplier (LM) test and the Schwarz Criterion (SC) [also know as the Schwarz Bayesian Information Criterion (SBIC)] were used to determine optimal lag length.  First, the LM-test was used to test for up-to-fourth-order serial correlation of the residuals to define the most parsimonious model with white-noise residuals.  Then, a few lags were added to determine the optimal lag length by minimizing the SC statistic.
 Phillips-Perron (1988) tests were also performed.  These tests involve a nonparametric correction for short-run dependence, rather than an autoregressive correction as in the ADF-test.
 Table 1 presents ADF and PP unit root tests of real foreign currency deposits, real consumption, inflation, black-market exchange rate, and shortage.  Inflation, pt, is used as the return on real assets. The official price index was used as deflator.  Each test was performed with a constant and no trend.  Since the residuals of the Phillips-Perron test do not have to be white-noise, a standard lag-length of four was used for this test.  MacKinnon (1991) provides response surface estimates of asymptotic critical values, which are printed below the t-tests in Table 1.  The significance levels reported next to each test statistic are small-sample significance levels computed by a bootstrap Monte Carlo simulation.  In Table 1, t-statistics indicate the null hypothesis of a unit root has to be accepted at the five percent critical level for all variables, except inflation in Hungary, with both tests based on the MacKinnon asymptotic critical values.

Table 1. Unit Root Tests
Country lags  ADF sig. level  lags PP sig. level
mtf 0 -2.48 0.99  -2.55 0.99
c 1 -0.24 1.00 4 -0.18 1.00
pt -1.32  0.96 -1.86  0.99
et 0.03  1.00 0.34 1.00
st 1 -0.96 1.00 4 -1.06  1.00
mtf 2 -0.76 1.00 4 -0.29 1.00
ct -0.21 1.00  0.73 1.00
pt -1.79 0.85  -6.87  0.26
et 1 -0.75 1.00 -0.84 1.00
st 1 -0.56 1.00  -0.93  1.00
mtf 4 -1.45 0.95 -1.27 1.00
ct -0.31 0.99 4 -0.34  1.00
pt 2 -2.24  0.94 4 -1.66  1.00
et 1 -1.49 1.00 4 -1.24 1.00
st 1 -1.30 1.00  4 -1.18 1.00
mtf 2.29 1.00 4 0.42 1.00
ct 3 -0.29  0.99 4 -1.23 1.00
pt 5 -0.07 0.99 -1.41 1.00
et 1 -0.27  1.00  3.67 1.00
st -0.68  1.00  4 -0.74  1.00
Critical Values 

The Phillips-Perron test for Hungarian inflation suggests it is I(0), although the ADF test suggests inflation is I(1).  Because the Phillips-Perron test is more robust to different choices of lag length, it would normally govern.  Monte Carlo evidence, however, shows that for all the variables tested, the small-sample distribution of both test statistics is shifted to the left, providing a bias toward rejecting the null hypothesis of an I(1) process.  When the Phillips-Perron test for Hungarian inflation is evaluated in the light of its small-sample distribution, the test is well within the region of acceptance of the null hypothesis of I(1).

Monte Carlo Simulations
 Asymptotic distribution theory often gives very misleading indications, so a Monte Carlo experiment was performed to examine small-sample properties of the test statistics for unit roots (ADF and Phillips-Perron (1988) statistics) and cointegration (Johansen maximum eigenvalue and Trace test statistics).  One thousand bootstrap Monte Carlo iterations were seeded with 90210573, using the empirical distribution of the actual data to simulate all variables for each of the four countries.  The bootstrap procedure is free of any assumptions about the data generating process.  In the bootstrap procedure, observations are randomly drawn with replacement, from the sample of actual observations.  Because observations are drawn with replacement, there is no limit to the number of iterations that can be performed.
 In each iteration, ADF and Phillips-Perron statistics were calculated and compared to the statistics calculated from the actual data.  Following Godfrey and Pesaran (1983), the percent of simulated test statistics more extreme than the actual test statistic approximates the small-sample marginal significance level of the actual.  This approximation approaches the true small-sample marginal significance level asymptotically.
 The ADF and Phillips-Perron tests are negative one-tailed tests with non-standard distributions; the percent simulated test statistics smaller than the actual are reported as the small-sample marginal significance levels in table 1.  The Johansen maximum eigenvalue and Trace tests are positive one-tailed tests with asymptotic distributions described by Johansen (1989) and Osterwald-Lenum (1992).

Long-run estimates
 Based on the derivation of the foreign money demand function in Section 2, we assume the following exponential specification:

(4.1)  m= e(ß0 + ß1p)cß2sß3xß4
Taking natural logs yields:
(4.2)   ln mf  = ß0 + ß1p + ß2 ln c + ß3 ln s + ß4 ln x,
where ß1 is the slope of the money demand function with respect to the inflation rate, and  ß2, ß3, ß4 represent elasticities of money demand with respect to consumption, shortage, and the black-market exchange-rate.
 Because the relationship to be estimated contains more than two variables, the long-run cointegration vector found by the conventional Engle-Granger (1987) two-step procedure or the single-equation error-correction model (Kremers et al, 1992) may or may not be unique.  Therefore, the maximum likelihood procedure of Johansen (Johansen (1988), Johansen and Juselius (1990)) seems a preferable method to test and estimate cointegration relationships.
 To apply the Johansen procedure, optimal lag length for the VAR was first determined by the likelihood ratio test advocated by Sims (1980).  The optimal lag length for Bulgaria and Poland was found to be 2, for Hungary and Romania, 4.  The trace test indicates  two cointegrating vectors for Bulgaria, two for Hungary, five for Poland,, and  three for Romania.
Table 2. Johansen Tests for Cointegration
Country (lags) r Trace (LR)* 5% crit. val. c2
Bulgaria (2) #4 0.695 3.76  35.63
#3  9.789 15.41
#2 28.47 29.68
#1 59.09 47.21
=0 117.1 68.52
Hungary (4) #4 0.082 3.76 144.12
#3 12.39 15.41
#2 27.65 29.68
#1 65.18 47.21
=0 127.21 68.52
Poland (2) #4 4.146 3.76 11.72
#3 17.86 15.41
#2 43.79 29.68
#1 94.40 47.21
=0 162.94 68.52
Romania (4) #4 5.66 3.76  156.58
#3 15.24 15.41
#2 70.79 29.68
#1 141.00 47.21
=0 290.22 68.52
* Critical values for the trace test are the 95% critical values given by Osterwald-Lenum (1992, table 1, p. 468).  The 99% critical value for the c2 test of restrictions on the ß matrix is 6.63.  This tests whether shortage can be deleted from the system of cointegrating vectors.  The trace test indicates two cointegrating vectors for Bulgaria, two for Hungary, five for Poland,, and  three for Romania.

 A Monte Carlo simulation with limited iterations indicated the Johansen procedure has superlative small-sample properties as a test for cointegration, and also has the desirable property of being extremely sensitive in distinguishing between cointegration among stationary variables and cointegration among non-stationary variables, a virtue not shared by either the augmented Engle-Granger or Kremers tests.
 Finally, Table 2 presents chi-square tests of restrictions on the ß matrix.  The restriction tested is that the shortage elasticities in each cointegrating vector are jointly equal to zero.  If the null hypothesis is accepted, shortage may be deleted from the system, and contributes no explanatory power.  The null is clearly rejected for all countries.
 Table 3 gives estimates of the foreign-money-demand elasticities and consumption functions for the four countries.   Because at least two cointegrating vectors were found for each country, the convention was adopted of reporting two vectors with the first normalized with respect to foreign-money-demand and the second with respect to real consumption spending.  Although the normalizations are arbitrary, it is convenient to interpret them as a demand function for foreign currency, and a consumption function.

Table 3. Cointegrating vectors
ln mft = a1 + b1pt + c1 ln st + d1 ln et
ln ct = a2 + b2pt + c2 ln st + d2 ln et
Country equation constant inflation shortage  exchange rate
Bulgaria mf 6.273 0.160 (0.128)  -0.392 (0.142)*  1.087 (0.161)*
consumption  8.362  -0.176 (0.030)* -0.422 (0.034)*  0.547 (0.038)*
Hungary mf -8.974 -2.361 (1.59) -4.636 (0.076)* 3.759 (0.165)*
consumption 5.657 -1.105 (0.522)* 0.159 (0.025)* 0.030 (0.054)
Poland mf 1.859  -0.00326 (0.0011)* -0.897 (0.368)*  1.085 (0.132)*
consumption -2.032 -0.00052 (0.0011)*  0.218 (0.367) 1.375 (0.131)*
Romania  mf -8.547 1.219 (2.375) -1.918 (0.286)* 2.350 (0.334)*
consumption 5.873  -3.416 (0.686)* -0.755 (0.083)* 0.765 (0.097)*
Standard errors are shown in parentheses.  Statistical significance at the 5% level is indicated with *.

 The consumption functions have negative coefficients for inflation, and positive coefficients for the exchange rate.  Inflation coefficients are statistically significant at conventional levels, except for Poland.  Exchange rate elasticities of consumption are significant except for Hungary.  An increase in the exchange rate indicates a decrease in the value of the domestic currency unit against the dollar.  Negative coefficients for inflation and exchange rates in the consumption function reflect the fact that consumption and income, expressed in real terms, fall as the value of the domestic currency unit falls.  Since the scope for black-market consumption in Bulgaria and Romania was small, official sector consumption should be interpreted as a proxy for income rather than actual consumption.  Foreign money predominated as a store of value in these economies rather than a black-market means of payment.  This is exactly what would be expected for countries with less-well-developed black markets.  Hungary and Poland had better-developed black markets early on.
 The shortage coefficients in the consumption functions are negative and significant for Bulgaria and Romania, which indicates as official sector shortage increased in severity, household consumption fell.  In contrast, consumption elasticities with respect to shortage are positive and significant for Hungary and positive but not significant for Poland, indicating official sector shortage brought about increased consumption spending.  This result could only have occurred with well developed, and highly active black markets, which were in fact operating in these economies well before the collapse of central planning.
 Inflation coefficients in the consumption functions should be unambiguously negative and significant because domestic inflation makes foreign money a better store of value, shifting consumption out of the official sector, and this is observed for all four countries.
 In the foreign money demand functions, the inflation coefficients are positive but not significant for Bulgaria and Romania, but negative and not significant for Hungary and negative and significant Poland.  Again, this reflects the different roles played by foreign currency holdings in transition economies with and without highly developed black markets.  In Bulgaria and Romania, where black markets were not well developed, foreign currency played predominantly a store-of-value role, and foreign currency holdings rose with high inflation.  In Hungary and Poland, with their highly-developed black markets, the transactions motive apparently predominated, and in response to inflation, households ran down their foreign currency balances in an effort to maintain steady consumption patterns.  This effect is seen most clearly in Poland, where the inflation coefficient is negative and significant.  Poland had the best-functioning black markets, and therefore the greatest scope for increased black-market transactions in response to high inflation, running down dollar balances held by Polish households.
 Shortage elasticities of foreign money demand are negative and significant for all countries.  This indicates that as shortage became more severe, households shifted consumption to the black market, running down foreign currency balances.
 The black-market exchange-rate elasticity of foreign money demand is positive and significant for all countries.  As with inflation, whenever domestic currency depreciates vis-à-vis foreign currency, people tend to substitute domestic money for foreign money.  As the black-market exchange-rate rose, foreign money became a better store of value.  Bias might be expected due to collinearity between inflation and the black-market exchange rate, but they were found to have very low correlation coefficients. In addition, as the exchange rate rose, foreign currency deposits expressed in terms of domestic currency went up even if the amount of foreign currency on deposit was constant, for obvious reasons.

5. Summary and Conclusion
 This paper examined currency substitution in four transition economies, Bulgaria, Hungary, Poland, and Romania, for 1986 to 1994.  Following liberalization of these economies in the beginning of the 1990s, currency substitution became an important phenomenon.  In most transition economies, currency substitution seems strongly correlated with high inflation and macroeconomic uncertainty.
 A formal choice-theoretic two-currency shortage-augmented cash-in-advance model was derived to account for both the period of central planning and liberalization of the economies.  Empirical estimates of the derived foreign-currency demand and consumption functions were performed by the Johansen procedure.  Results showed foreign currency deposits are explained very well by domestic official consumption, inflation, the black market exchange rate, and shortage.  The shortage variable, which could be interpreted as a permanent dummy for regime-shifts, was highly significant in explaining demand for foreign currency in all countries.  Recognizing the importance of the black market and currency substitution shows formal choice-theoretic modeling extremely powerful for analyzing money demand in these economies.
 The high degree of currency substitution has unambiguously negative effects for both fiscal and monetary policy in the transition economies.  First, currency substition reduces the scope for raising revenue through an inflation tax.  This implies that inflation will be higher, given a certain fiscal deficit.  Second, the high degree of currency substitution complicates monetary stabilization, one of the prerequisites for a successful transition to a market economy.  Third, very small changes in exchange-rate expectations might trigger large exchange-rate movements, due to substitution back and forth.  This further undermines the possibilities for monetary stabilization and use of the exchange rate as an anchor in the transition process.  Therefore, monetary authorities should attempt to reduce the degree of currency substitution in transition economies.
 Currency substitution tends to bring about higher rates of both depreciation of the domestic currency and inflation, than a transition economy would experience in its absence of currency substitution.  Currency substitution is self-enforcing because higher depreciation and inflation rates both contribute to even greater demand for foreign currency.  Taking into account its self-enforcing character, currency substitution can be extremely harmful to macroeconomic liberalization and stabilization programmes of the economies in transition.

Data appendix

All data sources used are official statistical publications of the four countries, the International Financial Statistics (IMF), Short-term Economic Indicators Central and Eastern Europe (OECD), and World Currency Yearbook (International Currency Analysis Inc, New York).  The latter publication has been used for black-market exchange rates. Most data on Bulgaria, Hungary, and Romania, were kindly provided by Bas van Aarle (see Aarle and Budina, 1995).  Where data were lacking, we supplemented his data set to create a homogenous data set for Bulgaria, Hungary, Poland and Romania for the period 1986.1 to 1994.1 (for Poland we extended the data set to 1994.2). The data on Poland were collected at the Money and Research Institute of the National Bank of Poland. The specific data sources are printed under each series.
Bulgaria-  Units: Billion Leva;  Sources: Aarle and Budina (1995) (BNB, except et (black market exchange rate) from World Currency Yearbook and  ct  (official-sector consumption) from OECD.)
Hungary-  Units: Billion Forints;  Sources: Aarle and Budina (1995) (IMF, NBH, except et  (black market exchange rate) from World Currency Yearbook and ct (official-sector consumption from OECD.)
Poland-  Units: Billion Zlotys;  Source:  Nijsse and Sterken (1995), ct (official-sector consumption) from OECD.)
Romania-  Units: Billion Leu;  Source:  Aarle and Budina (1995) (IMF, NBR, except et (black-market exchange rate) from World Currency Yearbook, and ct (official-sector consumption) from OECD.
Tables of the data are provided in Mulligan and Nijsse, 1995.


Aarle, Bas van (1994), "Currency Substitution and Currency Controls: the Polish experience of 1990", Tilburg University Faculty of Economics Research Memorandum No. 637.

Aarle, Bas van, Nina Budina (1995), "Currency Substitution in Eastern Europe", Center Discussion Paper No. 9502.

Bod, Péter (1994), "Targets and Instruments of Monetary Policy in Hungary", in: Tomás J.T. Baliño and Carlo Cottarelli (Eds.), Frameworks for Monetary Stability, Washington: International Monetary Fund.

Bulgarian National Bank (BNB), Annual Report, various issues, Budapest: BNB.

Calvo, Guillermo A., and M.S. Kumar (1993), "Financial Sector Reforms and Exchange Rate Arrangements in Eastern Europe", Occasional Paper, Washington: IMF.

Calvo, Guillermo A., and Carlos A. Végh (1992), "Currency Substitution in Developing Countries: An Introduction", Revista de Analisis Economico, 7, 3-29.

Charemza, Wojciech W., and Quandt, R.E. (1982), "Models and Estimation of Disequilibrium for Centrally Planned Economies, Review of Economic Studies, Vol. 49, pp. 109-116.

Charemza, W., and Ghatak, Subrata (1990), "Demand for Money in a Dual-Currency Quantity Constrained Economy: Hungary and Poland, 1956-1985", The Economic Journal, Vol. 100, pp. 1159-1172.

Charemza, Wojciech W. (1992), "East Europe in Wonderland, or: What happened to the money?",  University of Leicester Discussion Papers in European Economic Studies No. 92/8.

Charemza, W., and Gronicki, M. (1988), "Plans and Disequilibria in Centrally Planned Economies (Empirical Investigation for Poland)", Amsterdam: North-Holland.

Chawluk, A., and Cross, R. (1993), "Zloty and Dollar Balances in Poland, 1965-1993",

European Bank for Reconstruction and Development (EBRD) (1994), Transition Report, London: EBRD.

Engle, R.F., and Granger, C.W.J. (1987), "Cointegration and Error Correction: Representation, Estimation and Testing", Econometrica, 55, pp. 251-276.

Feltenstein, Andrew, David Lebow, and Sweder van Wijnbergen (1990), "Savings, Commodity Market Rationing, and the Real Rate of Interest in China", Journal of Money, Credit, and Banking, 22, 234-252.

Godfrey, L.G., and M.H. Pesaran (1983), "Tests of Non-nested Regression Models: Small Sample Adjustments and Monte Carlo Evidence", Journal of Econometrics, 21, 133-154.

Hartwig, K.-H. (1987), "Die Geldnachfrage in sozialistischen Planwirtschaften: Polen 1950-1984", Kredit und Kapital, No. 3, pp. 378-401.

Hartwig, K.-H. (1985), "Theoretische und empirische Ansätze zur Bestimmung der Geldnachfrage in sozialistischen Planwirtschaften", in: H.J. Thieme (Ed.), Geldtheorie: Entwicklung, Stand und Systemvergleichende Anwendung, Baden-Baden, pp. 235-253.

Hartwig, K.-H. (1983), "Involuntary Liquid Assets in Eastern Europe: Some Critical Remarks", Soviet Studies, Vol. 35(1), pp. 103-105.

International Monetary Fund (1995), International Financial Statistics (CD-ROM), Washington: IMF.

Hungarian Statistical Office (HSO), Statistical Yearbook, various issues, Budapest: HSO.

International Currency Analysis Inc., World Currency Yearbook, several issues, New York.

Jansen, P. (1982), Das Inflationsproblem in der Zentralverwaltungswirtschaft, Stuttgart: Fischer.

Jeffries, I. (1993), Socialist Economies and the Transition to the Market: A Guide, London: Routledge.

Johansen, Soren, and Juselius, Katarina (1990), "Maximum Likelihood Estimation and Inference on Cointegration - With Applications to the Demand for Money", Oxford Bulletin of Economics and Statistics

Johansen, Soren (1988), "Statistical Analysis of Cointegrating Vectors", Journal of Economic Dynamics and Control, Vol. 12, pp. 215-238.

Kornai, Janosz (1980), Economics of Shortage, Amsterdam: North-Holland.

Kremers, J.M.J., N.R. Ericsson, and J.J. Dolado (1992), "The Power of Cointegration Tests", Oxford Bulletin of Economics and Statistics, Vol. 54(3), pp. 325-346.

Lane, T.D. (1992), "Household Demand for Money in Poland", IMF Staff Papers, Vol. 39(4), pp. 825-854.

MacKinnon, James G. (1991), "Critical Values for Cointegration Tests," in R.F. Engle, C.W.J. Granger (eds.), Long-run Economic Relationships: Readings in Cointegration, Oxford: Oxford University Press.

Menger, Carl Friedrich (1871), Principles of Economics, English translation of Grundsaetze der Volkswirtschaftslehre, J. Dingwall and B.F. Hoselitz, trans., New York: New York University Press, 1976.

Mulligan, Robert F., and Erwin Nijsse (1995), "Currency Substitution in Transition Economies: Theory and Evidence," Kiel Institute of World Economics, mimeo.

National Bank of Hungary (NBH), Monthly Report, various issues, Budapest: NBH.

National Bank of Romania (NBR) (1994), Monthly Report, various issues, Bucarest: NBR.

Nijsse, Erwin, and Elmer Sterken (1995), "Shortages, Interest Rates, and the Demand for Money in Poland 1969-1994", University of Groningen, mimeo.

OECD, 1995, Short-term Economic Indicators Central and Eastern Europe (statistics on diskettes), January edition, Paris: OECD.

Osterwald-Lenum, Michael (1992), "A Note with Quantiles of the Asymptotic Distribution of the Maximum Likelihood Cointegration Rank Test Statistics", Oxford Bulletin of Economics and Statistics

Payne, James E. (1990), "A Model of Money Demand in Yugoslavia, 1952-1985", Economic Analysis and Workers Management, 24, 83-94.

Peebles, G. (1991), A Short History of Socialist Money, Sydney: Allen and Unwin.

Phillips, P.C.B., and P. Perron (1988), "Testing for a Unit-root in Time Series Regression", Biometrika, 75, 335-346.

Podkaminer, L. (1982), Estimates of Disequilibria in Poland's Consumer Markets, 1965-1978", Review of Economics and Statistics, Vol. 64(3), pp. 423-431.

Podkaminer, L. (1989), "Disequilibrium in Poland's Consumer Markets: Further Evidence on Intermarket Spillovers", Journal of Comparative Economics, Vol. 12, pp. 43-60.

Portes, R. (1981), "Macroeconomic Equilibrium and Disequilibrium in Centrally Planned Economies", Economic Inquiry, Vol. 9(9), pp. 559-578.

Portes, R., and Winter, D. (1980), "Disequilibrium Estimates for Consumption Goods Markets in Centrally Planned Economies", Review of Economic Studies, Vol. 47, pp. 137-159.

Rodriguez, Carlos Alfredo (1992), "Money and Credit Under Currency Substitution", IMF Working Paper No. 92.99.

Rojas-Suarez, Liliana (1992), "Currency Substitution in Peru", IMF Working Paper No. 92.33.

Sahay, Ratna, Carlos Végh (1995), "Dollarisierung in Transformationsländern", Finanzierung und Entwicklung (IWF und IBRD), 32, 34-37.

Savastano, Miguel A. (1992), "The Pattern of Currency Substitution in Latin America: An Overview", Revista de Analisis Economico, 7, 29-72.

Sims, C.A. (1980), "Macroeconomics and Reality", Econometrica, 48, 1-48.

Sturzenegger, Federico (1992), "Currency Substitution and the Regressivity of Inflationary Taxation", Revista de Analisis Economico, 7, 177-192.

Wyczanski, P. (1989), "The Dual Currency System in the Polish Economy and the Methods and Possibilities of its Reduction", Working Paper No. 3, National Bank of Poland Banking and Money Research Institute, Warsaw.