The Effects of Productivity Shocks, Financial Shocks, and Monetary Policy on Exchange Rates: An Application of the Currency Crisis Model and Implications for Emerging Market Crises

ABSTRACT

What kind of shock affects exchange rate dynamics? How much of an effect does the monetary policy have on exchange rates? To answer these questions empirically based on the currency crisis model, I use panel data on 51 emerging countries from 1980 to 2011, identify shocks, and apply instrumental variable methods. I found that both productivity shocks and shocks to a country’s risk premium affect exchange rates and a 1 percentage point increase in the policy interest rate is associated with a 1 percentage point appreciation of domestic currency. I further apply this method to Asian and Latin-American crises.

KEY WORDS:

• currency crisis
• exchange rates
• monetary policy
• productivity shock
• risk premium

JEL CLASSIFICATION:

• E4
• E5
• F3
• G15
• G2
• O47

Notes

^1. An interest rate defense is also important from macroprudential purposes (Nakatani 2016).

^2. See Bergman and Jellingsø (2010) for a proof of concavity and convexity of the two curves.

^3. Another way to analyze the effects of monetary policy shocks on exchange rates is to use VAR (Bernanke and Mihov 1998; Bjørnland 2009; Faust and Rogers 2003). However, this method is contentious because the results depend on identifying assumptions.

^4. A risk premium can be divided into the difference in nominal interest rates across currencies and the expected change in the exchange rate between these currencies. However, as Alvarez, Atkeson, and Kehoe (2009) argue, in the data, “the expected change in the exchange rate is roughly constant and interest differentials move approximately one-for-one with risk premia.” This is because exchange rates are roughly random walks (Cheung, Chinn, and Pascual 2005; Meese and Rogoff 1983) so that the expected depreciation of a currency is roughly constant and captured in the term ${\gamma }_1^i$ as a drift. Engel and West (2005) provide the theoretical justification for the random walk of exchange rates. Under some empirically plausible circumstances (if at least one of the underlying fundamentals has a unit root and the discount factor is near one), exchange rates are near-random walks. Engel, Mark, and West (2007) found that the forecasting ability of a random walk outperforms that of economic predictors when the models are estimated country by country. Rossi (2013) recently surveyed a broad range of literature up to date and concluded that “Messe and Rogoff’s finding does not seem to be entirely and convincingly overturned.”

^5. The smoothing parameter is set as 100 for annual data. The results do not change substantially when we alter the parameter.

^6. The IPLM shocks are considered idiosyncratic.

^7. Levin and Lin (1992) proposed another panel unit root test under the assumption of i.i.d. disturbances. However, O’Connell (1998) showed that the Levin–Lin test statistic is no longer correct when there is a cross-sectional heterogeneity. The Im–Pesaran–Shin test is generally better than the Levin–Lin test.

^8. Windmeijer (2005) found that the efficient two-step GMM estimator outperforms somewhat the one-step GMM estimator in estimating coefficients with lower bias.

^9. When the coefficient on AR(1) term is close to unity, it is known that the system GMM estimator that is proposed by Arellano and Bover (1995) and Blundell and Bond (1998) performs well, though this is not the case here since the coefficient is about 0.1 and far from unity.

^10. Note that this result does not guarantee the effectiveness of monetary tightening on exchange rates at high frequency during the crisis periods.

^11. Using data on three open economies, Zettelmeyer (2004) found similar results; a 1 percentage point increase in the policy interest rate is associated with a 2−3 percentage point appreciation of domestic currency.

^12. Note that the model presented in Calvo and Mendoza (1996) is not a pure second-generation model of currency crises and includes the features of the first-generation models since the level of international reserves plays a critical role for the collapse of the exchange rate regime.

^13. More than 80% of Argentina’s government debt was denominated in dollars by late 2001 (Edwards 2002).