Euroization: what factors drive its persistence? Household data evidence for Croatia, Slovenia and Slovakia

Abstract

The question asked in this article is why people continue to use foreign currencies even after their economies have stabilized. Survey data for Croatia, Slovenia and Slovakia are employed to provide an answer. The results confirm the role of network effects and of remittances. Furthermore, the extent of currency substitution is found to be positively associated with the level of income and education. An important aspect of euroization seems to be age (older people are more likely to hold foreign currencies). In contrast, neither expectations about inflation rates, nor about exchange rates, do seem to affect the degree of euroization in a systematic and predictable way. Trust in the banking system is found to affect the choice between foreign currency cash and foreign currency deposits. Overall, the results support the view that the persistence in the use of foreign currencies is driven to a large extent by factors that are related to the past.

Acknowledgements

I would like to thank the anonymous referees and Doris Ritzberger-Grünwald for valuable comments and discussions. The views expressed in this article are strictly those of the author and do not necessarily reflect the views of the Oesterreichische Nationalbank.

Notes

¹ Although the economic substance is the same, we use the term euroization instead of dollarization because the euro and its legacy currencies have had a predominant role in the countries analysed.

² For those Central, Eastern and South-Eastern European countries that are highly euroized, Feige and Dean (2004) observe that ‘… for network externality reasons the use of foreign currency […] for transactions purpose is unlikely to be reversible, even if they pursue moderate macroeconomic policies and hence reduce inflation risk’ (p. 21).

³ Related, Kyriakos and Savva (2006) emphasize the importance of macroeconomic stability in general.

⁴ Seater (2008) stipulates that the level of income plays an important role for euroization. See the discussion below. Compare also with the discussion on the income and wealth elasticity of money demand in dollarized economies (Adam et al., 2004).

⁵ We will use the following terminology in this article: currency substitution will refer to the substitution of foreign currency (cash, FCC) for domestic currency (cash, LCC). In general, asset substitution refers to the substitution of foreign denominated monetary assets to domestic denominated monetary assets. As we will not focus on other monetary assets apart from saving deposits, the term asset substitution and deposit substitution will be used interchangeably, both referring to the substitution of saving deposits denominated in foreign currency (FCD) for saving deposits denominated in local currency (LCD). Hence, euroization refers to the overall, and in our case, unofficial extent of currency and asset substitution. This terminology corresponds very closely to the terminology proposed by Feige et al. (2003).

⁶ Inflation rates were around 6–8% in Croatia and around 9% in Slovenia in 2000 and 2001. Slovakia had an inflation rate of 16% in 2000. During the year 2006, inflation rates ranged from only 2 to 4%.

⁷ The surveys were also conducted in Hungary and the Czech Republic. For the purpose of this study, we restrict our attention to Croatia, Slovenia and Slovakia because the number of individual observations for which we can identify that currency is held for reserve purposes and not to cover transactions made abroad (e.g. for trips) is low in Hungary and the Czech Republic.

⁸ Before the introduction of euro banknotes and coins, the German mark was the most frequently held foreign currency in the region. In Croatia, the US dollar and in Slovenia and Slovakia the Austrian schilling were the second-most important foreign currencies.

⁹ It should be noted that the chosen specification is likely to represent only an approximation to the demand for FCC and FCD, in particular as the relationship between the dependent and the independent variables can potentially be complex and highly nonlinear (cf. Seater, 2008).

¹⁰ For each survey and in each country approximately 1000 persons above the age of 14 years were personally interviewed in April/May and in October/November, such that our joint sample comprises four survey waves with about 4000 observations per country.

¹¹ Detailed information about the questionnaire as well as descriptive statistics are summarized in Appendix A.

¹² Balances associated with this motive are driven by very different economic considerations and are higher in value than balances associated with transactions abroad (cf. Ritzberger-Grünwald and Stix, 2007).

¹³ Due to data constraints, we must neglect the role of other financial assets. According to the surveys, 40% of Croats, 69% of Slovaks and 87% of Slovenes held a savings account. The particularly high value for Slovenia can be traced to an imprecise translation in the questionnaire because the English questionnaire referred to pure savings accounts while the Slovenian translation referred to both transaction (checking) and savings accounts.

¹⁴ Thus, the data suggest that foreign cash holdings are not necessarily perfect substitutes for foreign currency deposits. On the one hand, people could hold some FCC at home for liquidity reasons. On the other hand, respondents might not always correctly distinguish between cash and deposits.

¹⁵ The working article version of this article contains a third dependent variable which measures whether an individual holds foreign currency cash and/or foreign currency deposits.

¹⁶ Notice that the exchange rate expectations refer to the expectations vis-à-vis the euro while currency ownership entails both euro and US dollar holdings. Given that US dollar ownership rates are low, we consider this only a negligible inconsistency.

¹⁷ For example, Honohan (2007) or, for Croatia, Šošić and Kraft (2006), provide evidence that agents react to exchange rate changes by changing the currency composition of their deposits. However, their evidence refers to the direct impact via current and not expected exchange rate changes.

¹⁸ Which might not be a serious drawback as this approach strongly underpredicts observed dollarization levels in Croatia (Šošić and Kraft, 2006).

¹⁹ I thank a referee for pointing this out.

²⁰ This would be consistent with rational behaviour if, according to Seater (2008), additional interest earned with higher income dominates additional conversion and fixed costs (costs of conversion, shoe-leather costs, other costs associated with the deposit).

²¹ Results from a later survey from May 2006 show that if foreign currencies are used for payments, they are typically used for large value payments in Croatia and Slovenia. Only in Slovakia, answers indicate that small value payments dominate.

²² Many Croats have relatives in non-euro area countries (Switzerland, Australia, the US, Chile and Argentina). The focus on relatives from the euro area neglects the role of remittances from these countries. I thank a referee for pointing this out.

²³ Distances to nearest Austrian cities can be low: Bratislava to Vienna: 60 km, Ljubljana to Klagenfurt: 80 km; Zagreb to Graz: 180 km.

²⁴ The results of all specifications are reported in the working paper version of this article or available from the author upon request.

²⁵ Colacelli and Blackburn (2009), though analysing a different question, namely the likelihood of acceptance of a secondary currency, also report an effect of occupational dummies. Their interpretation is that the occupational dummies measure the extent of skill formation or human capital and hence of matching skills. In our case, we consider it more likely that the effect of occupation is caused by differences in the currency denomination structure of income.

²⁶ However, at the same time, our finding is somewhat in contrast to conjectures raised in the literature that network effects should only play a role for currency substitution but not for asset substitution. However, we consider the finding of network effects for deposits plausible as deposits are close substitutes for cash, in particular in the analysed countries which have well-developed financial systems.

²⁷ This is also confirmed by joint tests, revealing that the marginal effect is different for those between 25 and 54 years than for those above the age of 54 years. The test statistics is significant at a 1% level for Croatia and Slovakia and at a 10% level for Slovenia.

²⁸ The fact that the point estimates of the marginal effect for those Croats who consider deposits at banks very unsafe (−12) is larger than those who consider deposits rather safe (−18) might seem odd. However, the two effects are not different in a statistical sense. Furthermore, this effect might be traced to the low number of respondents for those considering deposits at banks very unsafe.

²⁹ We are left with the enigma that in some cases, currency substitution and asset substitution are actually negatively related over time. One provisional explanation of this phenomenon can be found in improvements in the domestic banking system…’ (Feige 2003, p. 27).