The role of monetary shocks and real shocks on the current account, the terms of trade and the real exchange rate dynamics: a SVAR analysis

Abstract

The article is an empirical investigation of the role of monetary shocks and real shocks on the current account, the terms of trade and the real exchange rate dynamics. Three new open economy macro-economics (NOEM) models are studied to motivate the empirical investigation. Then I apply the Blanchard–Quah decomposition to identify a SVAR model. The empirical evidence supports the NOEM model with the simplified household preference specification and with the small degree of pricing-to-market. The evidence also illustrates monetary shocks important for the real exchange rate and the terms of trade movements, but not for the current account fluctuations.

Acknowledgement

This article is revised from the second chapter of my PhD thesis completed at University of Virginia. I thank my dissertation advisors Dr. Eric van Wincoop, Chris Otrok, and Jinill Kim for helpful discussions. All remaining errors are the author's. I gratefully acknowledge helpful comments from an anonymous referee and Editor Taylor. The funding from Bankard Foundation and FERIS Foundation of America is appreciated.

Notes

¹ The policy evaluation and welfare analysis based on NOEM models also usually depends on the particular specification of preferences.

² Mankiw and Summers (1986) find an estimation of the consumption elasticity of money demand to be a number close to one. Betts and Devereux (1996), Hau (2000) also suggest a unit money demand elasticity in their papers.

³ It is necessary to define B_{t +1} as a nominal indexed bond in this pricing-to-market model instead of a real indexed bond in the baseline model because with the pricing-to-market feature, the traded goods prices are no longer equated across countries.

⁴ With a zero pass-through, the real exchange rate does not respond to the monetary shock; with a full pass-through, the current account stays in balance.

⁵ The only exception is that under a zero pass-through, the real exchange rate remains constant.

⁶ Iscan (2000) finds that the country-specific productivity growth in tradables and nontradables has different implications for the current account dynamics.

⁷ The third possibility is to have C ₁₁, C ₁₂ and C ₂₃ be zeros.

Notice that here, the C1 * (C1)′ is symmetric with two zeros.

With six unknowns but only five equations, it is impossible to solve C1 from C1 * (C1)′.

⁸ The Bayes information criterion (BIC) was also tried and found to suggest the same lag length as the AIC.

⁹ The idea behind it is that we get the small-sample distribution of impulse response functions without assuming that the errors in the reduced-form regression are Gaussian distributed. We first estimated the SVAR and saved the coefficient estimates and the fitted residuals {û _t }. These fitted residuals would be resampled with a replacement of . Based on the resampled residuals, we generated the values of . A SVAR would be fitted by OLS to the simulated . From this estimate, the coefficients of were calculated. Next, I generated a second set of the fitted residuals and the simulated and . A series of 1000 such simulations were undertaken, and a 95% confidence interval was inferred from the range that included 95% of the values for C _s .

¹⁰ The details of the solution are provided in Zhang (2006).

¹¹ Define to be the percentage deviation of the period 1 variable from the initial steady state in period 0, i.e. ; define to be the percentage deviation of the long-run variable from the initial steady state, i.e. .

¹² A negative implies a positive productivity shock to the tradable sector.