Official dollarization in El Salvador and the inflation–inflation uncertainty nexus

Abstract

This study extends the literature on the relationship between inflation and inflation uncertainty by examining the impact of official dollarization on inflation uncertainty in El Salvador. The ARIMA–GARCH model reveals that official dollarization significantly reduces the degree of volatility persistence in response to inflationary shocks. However, based on Granger causality tests, if inflation should increase there would be an increase in inflation uncertainty as suggested by the Friedman–Ball hypothesis.

Notes

¹The advantage of ARCH–GARCH models is that they only focus on the unpredictable component of inflation.

²Majority of the studies on inflation and inflation uncertainty using the ARCH–GARCH approach pertain to industrialized economies: Engle (1983), Bollerslev (1986), Cosimano and Jensen (1988), Evans (1991), Brunner and Hess (1993), Caporale and McKiernan (1997), Balcombe (1999), Hwang (2001) and Caporale and Caporale (2002) for the USA; Fountas (2001) and Kontonikas (2004) for the UK; Bohara and Sauer (1994) for Germany; Baillie et al. (1996), Grier and Perry (1998), Berument and Dincer (2005) and Henry et al. (2007) for the G7 countries; Fountas et al. (2004) for European Union countries; Conrad and Karanasos (2005) for the USA, the UK and Japan; and Berument and Yuksel (2007) and Payne (2008b) for inflation-targeting countries. Baillie et al. (1996), Nas and Perry (2000), Telatar and Telatar (2003), Daal et al. (2005), Berument and Yuksel (2007), Thornton (2007, 2008) and Payne (2008a) have examined the developing and the emerging market economies.

³Dummy variable for the official dollarization period was initially included in the mean equation but was statistically insignificant and subsequently omitted from the equation.

⁴As a check on the robustness of the Granger causality tests, lagged inflation was included in the conditional variance equation and the conditional standard deviation in the mean equation. Lagged inflation was positive and statistically significant at the 1% level while the conditional standard deviation was statistically insignificant. Results are available upon request.