Portfolio flows and the US dollar–yen exchange rate

This paper investigates the effects of portfolio flows on the US dollar– Japanese yen exchange rate changes over the period 1988:01–2011:04. Using a time-varying transition probability Markov-switching framework, the results suggest that the impact of portfolio flows on the dollar–yen exchange rate changes is statedependent. In particular, the results show that portfolio inflows from Japan toward the US, more than monetary variables, strengthen the probability of remaining in the dollar–yen appreciation (low volatility) state. Therefore, credit controls on the flows can be used as a policy tool to pursue economic and financial stability.


Introduction
Over the recent years there has been a signi…cant attention on the cross-border portfolio ‡ows e¤ects on exchange rate changes and its volatility. The general view is that large in ‡ows result in an exchange rate appreciation. Nonetheless, volatile cross-border ‡ows may lead to an increase in the volatility of the exchange rates. In this paper, we examine to what extent equity and bond portfolio ‡ows between the US and Japan a¤ect the corresponding US dollar-yen exchange rate changes, given that the cross-border acquisition of long-term securities between these countries has grown over the recent years (see Figure 1). The US listed stocks of Japanese companies have been substantially larger compared to the other developed countries. Also, the Bank of Japan has frequently exercised what is known as the sterilized intervention in the foreign exchange market using long-term, primarily US, bond instruments to tackle the stagnation of the Japanese economy. 1 As Sarno et al. (2015) documented, bond ‡ows between the US and Japan has also been driven by the so-called yen carry trade, where investors borrowed in yen at the very low rates to invest in high interest currencies, mainly the US dollar. 2 The lack of academic consensus on what factors explain the dollar-yen exchange rate changes further motivates this study. Obstfeld (2009)  Considering the recent evidence of Menla Ali et al. (2015) on nonlinear dependence, this paper uses the time varying transition probability Markov switching framework as an alternative way to examine the nonlinear impact of the ‡ows on the dollar-yen exchange rate changes. The adopted framework is ‡exible enough to capture the nonlinearity in the relationship between exchange rate changes and portfolio ‡ows as it separates periods of appreciation from periods of depreciation and of high volatility from those of low volatility allowing the probabilistic structure of the transition from one regime to the next be a function of the ‡ows. The causal e¤ect is not constrained to be symmetric in the parameterization (portfolio ‡ows a¤ect the exchange rate di¤erently in periods of appreciation and depreciation, as well as high and low volatility) and in the temporal causality (portfolio ‡ows have a di¤erent impact on the future exchange rate in periods of appreciation and depreciation, as well as high and low volatility). The model, therefore, measures the impact of portfolio ‡ows for di¤erent states of the currency. 3 Intuitively, portfolio ‡ows may not be the same during periods of appreciation and depreciation and also when the currency is highly volatile and less volatile. In fact, there is now evidence that international ‡ows change with the level of uncertainty in the foreign exchange markets. Caporale et al. (2015) found speci…cally that international in ‡ows towards the US were dampened during high uncertainty of exchange rate changes using data from major developed countries. Figure 2 displays the evolution of the net portolio in ‡ows from Japan towards the US and the annualized historical volatility of the dollar-yen exchange rate changes. A graphical inspection suggests that periods associated with large increases (declines) in the in ‡ows correspond to those of relatively low (high) volatility of exchange rate changes.
[Insert Figures 1 and 2 about here] The adopted nonlinear analysis is further of paramount interest, given that the safe haven feature of the yen during certain periods, such as the period of the global …nancial crisis, has been documented by several studies. Habib and Stracca (2012) reported that safe haven currencies, per se, tend to have a stronger net foreign asset position. It follows that portfolio ‡ows'behaviour is likely to be di¤erent during periods when the currency acts as a safe haven. 4 The nonlinear mechanism adopted in this paper o¤ers important policy implications. Policymakers and regulators can set appropriate policies with regard to economic and …nancial stability.
For example, policies aimed to ease the impact of currency appreciation and the associated negative e¤ects on the economy could be unsuccessful if higher in ‡ows move the exchange rate to the appreciation regime. Credit controls on the ‡ows may be deemed in this regard, since such a policy is likely to mute the in ‡ows, and hence stabilize the economy.
The remainder of this paper is organized as follows. Section 2 outlines the econometric model and the hypotheses tested in this study. Section 3 describes the data and discusses the empirical results, and Section 4 concludes.

The model
The time varying regime-switching model considered in this paper allows for shifts in meanvariance, that is for periods of appreciation and depreciation and low and high volatility, and is given by: where r t denotes the log changes of the US dollar-Japanese yen exchange rate. Autoregressive terms (up to four lags) are also considered. Therefore, the parameters vector of the mean equation, (Eq. 1); is de…ned by (i) (i = 1; 2) and (i) (i = 1; 2) which are real constants, the autoregressive errors with E(" t ) = 0 and E(" 2 t ) = 1, and the random variables fs t g in S = f1; 2g which indicate the unobserved state of the process at time t. Throughout, the regime indicators fs t g are assumed to form a Markov chain on S with transition probability each column sums to unity and all elements are nonnegative. It is also assumed that f" t g and fs t g are independent.
To assess the links between net portfolio ‡ows and the dollar-yen exchange rate changes, we generalize the model in Eq. (1) by allowing the transition probabilities to vary over time. In particular, we assume that each conditional mean follows an independent regime-shifting process and, following Filardo (1994), the transition mechanism governing fs t g is given by (Model 1): ; Note that since p 1 t =npf t 1 has the same sign as 1 , 1 > 0 implies that an increase in net portfolio in ‡ows, npf t 1 ; increases the probability of remaining in state 1. Similarly, 1 > 0 implies that an increase in npf t 1 will increase the probability of remaining in the second regime. 5 For robustness purposes, the following control variables are considered: stock market return di¤erential (s s ) t 1 , short-term interest rate di¤erential (i i ) t 1 and real oil price changes oil t 1 . 6 (s s ) t 1 captures stock market shocks across the US and Japan. (i i ) t 1 controls for the di¤erent monetary policies between the US and Japan as a result of the di¤erent in ‡ationary environments over the period under investigation (see Bernanke, 2000). oil t 1 is aimed to capture the terms of trade shocks (Amano and van Norden, 1998) as both the US and Japan are net importer countries and the input costs in the two countries are highly sensitive to oil price changes.
Therefore, the extended model (Model 2) has the following form: ; Finally, we take as a benchmark the standard linear model frequently estimated in the literature (see, Brooks et al., 2004;Hau and Rey, 2006), and speci…ed as follows: Further details on the estimation process and the employed data are given in the next Section.

Data Description and Empirical results
The Exchange rate changes are calculated as r t = 100(E t =E t 1 ); where E t denotes the log exchange rate at time t. Net portfolio ‡ows, by contrast, are constructed as the di¤erence between portfolio in ‡ows and out ‡ows. In ‡ows and out ‡ows are measured as net purchases and sales of domestic assets (equities and bonds) by foreign residents, and net purchases and sales of foreign assets 5 Note that failure to reject the null hypothesis of H0 : 1 = 1 = 0 suggests a …xed transition probabilities (FTP) model. 6 The asterisk refers to Japan. 7 For a detailed description of the TIC data, see Edison and Warnock (2008 [Insert Table 1 and Figure 3 about here] The null hypothesis of linearity against the alternative of a Markov regime switching process cannot be tested directly using a standard likelihood ratio (LR) test. We properly test for multiple equilibria (more than one regime) against linearity using Hansen's (1992) test. The results (Table   1) support a two-state regime-switching model. The presence of a third state was tested and rejected.
Model 1 appears to be well identi…ed (see Table 1). The standardized residuals exhibit no signs of linear or nonlinear dependence. The periods of appreciation and depreciation seem to be accurately selected by the smoothed probabilities, which clearly separates the two regimes.

Conclusion
In this paper, we have investigated the causality dynamics running from portfolio in ‡ows into the US dollar-yen exchange rate changes using quarterly data over the period 1988:01-2011:04. By The results presented herein are robust to monetary policy, stock markets and terms of trade shocks. Furthermore, they add new information suggesting that portfolio in ‡ows rather than monetary variables keep the domestic currency in the appreciation regime. The dynamic time varying approach supports the appropriateness of the nonlinear framework adopted in this paper.
The policy implication of our result is that portfolio ‡ows are more e¢ cacious in determining the currency dynamics, and therefore credit controls on these ‡ows can be used in order to pursue economic and …nancial stability.      Notes: a r t , npf t ; (s s ) t ; (i i ) t ; and oil t indicate exchange rate changes, net portfolio ‡ows, equity return di¤erential, interest rate di¤erential, and real oil price changes, respectively. JB is the Jarque-Bera test for Eq. (3), respectively. Autocorrelation and heteroscedasticity-consistent standard errors are used and represented in parentheses (.). LB (8) and LB 2 (8) are respectively the Ljung and Box (1978) tests of signi…cance of autocorrelations of eight lags on the standardized and squared standardized residuals. P -values are reported in square brackets [.].