Measuring the speed of currency substitution

ABSTRACT

Financial dollarization is a prevalent and persistent phenomenon among emerging and developing economies. This paper uses data on deposit dollarization in 173 countries between 1975 and 2018 to present stylized facts on the speed of currency substitution, which has not previously been systematically measured. In doing so, the paper casts a wide net in defining what constitutes a currency substitution episode. Overall, when currency substitution happens, it happens fast: the share of total deposits that is denominated in foreign currency rises by 6.5 percentage points per year during an average episode.

KEYWORDS:

• Dollarization
• de-dollarization
• currency substitution
• foreign currency deposits

JEL CLASSIFICATION:

• F31
• G15

Acknowledgments

I am grateful to Soledad Martinez Peria, Martin Čihák, Anita Tuladhar and Yiqun Wu for comments and to Chenxu Fu for excellent research assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Disclaimer

The views expressed herein are those of the author and should not be attributed to the IMF, its Executive Board, or its management.

Notes

¹ The use of a foreign currency as a store of value in the financial sector is normally also a prerequisite for its use as a medium of exchange: financial dollarization is an important driver of the dollarization of retail payments (Drenik and Perez 2021). A common threshold for high financial dollarization is that the share of foreign currency deposits in total bank deposits is over 30% (Baliño, Bennett, and Borensztein 1999).

² The risks and implications of currency substitution are coming under renewed scrutiny due to the rise of digital currencies. To the extent that private cryptocurrencies or digital currencies issued by foreign central banks will spur foreign currency adoption in EMDEs, a new wave of (digital) dollarization could emerge (IMF 2020, 2021).

³ The reason that this panel is unbalanced panel is that not all countries have data entries for all years (in particular, many start later than 1975). We have observations on 4,234 country-year pairs out of a maximum possible of 7,612 (i.e. for a balanced panel from 1975 till 2018 with 173 countries).

⁴ In a few cases, we supplement this SRF data with historical data from Bannister, Gardberg, and Turunen (2018) or Levy Yeyati (2006), to obtain series that go back farther in time. We have checked that merging this data does not lead to time-series jumps. We further note that the data do not allow for a currency breakdown (i.e. per country, deposits for all foreign currencies are grouped together). As is common in the literature, this paper uses the term ‘dollarization’ as a synonym for ‘currency substitution’, rather than specifically substitution into US dollars.

⁵ Episode identification runs from the last year below the lower threshold to the first year above the upper threshold, regardless of variation between the thresholds. For example, a dollarization pattern like ‘year 1 = 18%’, ‘year 2 = 31%’, ‘year 3 = 22%’, ‘year 4 = 41%’ would be counted as a single episode in the row that measures dollarization from below 20 to above 40% in Table 1.