Trilemma, dilemma or 2.5-lemma in the transmission of monetary policy: evidence from a Markov-switching panel data model

ABSTRACT

We examine the spill-over effects of interest rate transmission of United States monetary policy to peripheral countries with various exchange rate regimes and capital control management policies. To do so, we propose a two-state continuous-time hidden Markov-switching panel data model using an empirical framework based on the Taylor rule. We find peripheral countries with flexible exchange rates and free capital mobility respond differently to changes in short-term US interest rates under the two regimes based on interest rate volatility. Under the high volatility regime, peripheral countries respond to decreases but not to increases in short-term US interest rate. Under the low volatility regime, peripheral countries respond more strongly to an increase than a decrease in short-term US interest rate. In addition, peripheral countries under high volatility regimes respond more strongly than countries under low volatility regimes to a decrease in short-term US interest rate. In both volatility regimes, capital controls insulate the peripheral countries from changes in the short-term US interest rate regardless their exchange rate regimes.

KEYWORDS:

• Panel Markov switching
• exchange rate regimes
• capital control
• trilemma
• dilemma

JEL CLASSIFICATION:

• E42
• E52

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Aizenman, Chinn, and Ito (2016).

² We conduct a Wu-Hausman test for endogeneity of $\mathrm{\Delta }GD{P}_{k,t}$ and $\mathrm{\Delta }In{f}_{k,t}$on Equation (5)(5) $\begin{array}{rl}& {\mu }_{k,t}^{(i)}={\beta }_0^{(i)}+{\beta }_1^{(i)}{r}_{k,t-\delta }+{\beta }_2^{(i)}\mathrm{\Delta }GD{P}_{k,t}+{\beta }_3^{(i)}\mathrm{\Delta }In{f}_{k,t}\\ & \hspace{0.17em}+{\beta }_4^{(i)}{D}_{fixed,NC}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_5^{(i)}{D}_{fixed,C}\mathrm{\Delta }{r}_{k,t}^{US}\\ & +{\beta }_6^{(i)}{D}_{flex,NC}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_7^{(i)}{D}_{flex,C}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_8^{(i)}\mathrm{\Delta }VI{X}_t\end{array}$(5) using the method proposed by Lewbel (2012) with the ‘REndo’ package in R. The test results provide support for evidence that $\mathrm{\Delta }GD{P}_{k,t}$ and $\mathrm{\Delta }In{f}_{k,t}$are not endogenous. The test details and the results of the endogeneity tests are available from the authors upon request.

³ Han and Wei extend the model to include the change in real exchange rate for 1999 to 2009. Using the same data, we find that the change in real exchange rate is significantly correlated with the change in inflation at −0.99. We also include the change in real exchange rate in our benchmark one-state regressions with results unaffected by multicollinearity. However, the results of 2-state HMM regressions on coefficients of $\mathrm{\Delta }In{f}_{k,t}$ and $\mathrm{\Delta }GD{P}_{k,t}$show indications of multicollinearity. The results of the other coefficients of the responses to $\mathrm{\Delta }{r}_{k,t}^{US}$ (${\beta }_4\mathrm{t}\mathrm{o}{\beta }_7$ in Equation 5(5) $\begin{array}{rl}& {\mu }_{k,t}^{(i)}={\beta }_0^{(i)}+{\beta }_1^{(i)}{r}_{k,t-\delta }+{\beta }_2^{(i)}\mathrm{\Delta }GD{P}_{k,t}+{\beta }_3^{(i)}\mathrm{\Delta }In{f}_{k,t}\\ & \hspace{0.17em}+{\beta }_4^{(i)}{D}_{fixed,NC}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_5^{(i)}{D}_{fixed,C}\mathrm{\Delta }{r}_{k,t}^{US}\\ & +{\beta }_6^{(i)}{D}_{flex,NC}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_7^{(i)}{D}_{flex,C}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_8^{(i)}\mathrm{\Delta }VI{X}_t\end{array}$(5) and ${\beta }_4\mathrm{t}\mathrm{o}{\beta }_9$in Equation 6)(6) $\begin{array}{rl}& {\mu }_{k,t}^{(i)}={\beta }_0^{(i)}+{\beta }_1^{(i)}{r}_{k,t-\delta }+{\beta }_2^{(i)}\mathrm{\Delta }GD{P}_{k,t}\\ & +{\beta }_3^{(i)}\mathrm{\Delta }In{f}_{k,t}\hspace{0.17em}+{\beta }_4^{(i)}{D}_{fixed,NC}\mathrm{\Delta }{r}_{k,t}^{US}\\ & +{\beta }_5^{(i)}{D}_{fixed,C}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_6^{(i)}{D}_{flex,NC,USpos}\mathrm{\Delta }{r}_{k,t}^{US}\\ & +{\beta }_7^{(i)}{D}_{flex,NC,USneg}\mathrm{\Delta }{r}_{k,t}^{US}+{\beta }_8^{(i)}{D}_{flex,C,USpos}\mathrm{\Delta }{r}_{k,t}^{US}\\ & +{\beta }_9^{(i)}{D}_{flex,C,USneg}\mathrm{\Delta }{r}_{k,t}^{US}\hspace{0.17em}+{\beta }_{10}^{(i)}\mathrm{\Delta }VI{X}_t.\end{array}$(6) are consistent with or without the change in real exchange rate as regressor.

⁴ The data are available at Xuehui Han’s website at https://drive.google.com/file/d/1DZ5h6BcWNCVL5bh5ffHG-Z-nTlWYg7X1/view. Briefly, the data includes the GDP and CPI projections from the IMF. the short-term interest rates from the International Financial Statistics and the long-term interest rates from CEIC, Haver and Investing.com. Han and Wei discuss the details of the data and descriptions of exchange rate and capital control regimes in an online Appendix.

⁵ Our results for the 1-state model replicate Han and Wei’s results but our criteria are significance at 5% and higher, while Han and Wei’s are at 10% and higher.

⁶ As in Han and Wei (2018), we don’t consider asymmetric effects on peripheral countries with fixed exchange rate regimes because there are too few observations within the sub-categories with and without capital controls.

⁷ The Wu-Xia shadow federal funds rate data are available at https://www.frbatlanta.org/cqer/research/wu-xia-shadow-federal-funds-rate

⁸ We conducted robustness checks for alternative measures of capital controls by Klein’s (2012), Chinn and Ito (2008) and Fernández et al.’s (2015) and of exchange rate regimes by the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER) and by Shambaugh (2004). Our findings on the short term interest rates are robust to alternative specifications of capital controls and exchange rate regimes but less robust on the low volatility long-term interest rate state. The robustness checks are available from the authors upon request.

Additional information

Funding

This work was supported by the Tier 1 Ministry of Education, Singapore [grant reference RG169/18]