The Market Dynamics: Time Series Analysis of Capital Flows and Exchange Rates Open Access
Downloadable ContentDownload PDF
Major financial crises in the past decades were associated with either extreme capital flows or extreme exchange rate movements, or both. The existence of crises hints that capital flows and exchange rates may not evolve in a linear way, but rather reflect the complex dynamics in the international financial market. This dissertation examines the behaviors of capital flows and exchange rates using nonlinear time series methods. Regime dependence, bubble behavior, and country heterogeneity are taken into consideration to both provide new empirical facts and motivate future theoretical advancement.Chapter 1 investigates the possible existence and nature of regime-shifting nonlinear dynamics in gross capital flows. Gross capital flows are much larger and more volatile than net flows, thus more relevant for financial stability issues. This chapter tests for nonlinearity in gross capital flow dynamics and finds that nonlinearity commonly exists in both total inflow and outflows. The dynamics are then analyzed in a threshold autoregressive model (TAR) which augments the standard AR model with a nonlinear term capturing different economic states. Furthermore, conditional heteroskedasticity is considered to account for possible volatility clustering commonly observed in financial data. The estimation results show that the dynamics of gross flows are different in crisis times and normal times; the dynamics are more unstable in crisis times especially for relatively risky countries. In addition, a generalized impulse response function (GIRF) analysis is conducted, allowing regime-dependent responses to a capital flow shock. The results show that a capital flow shock tends to be larger and more persistent in a crisis regime. This chapter suggests that the dynamics of gross capital flows may be regime-dependent and the impacts of capital flow shocks should be evaluated conditional on economic situations and capital flow patterns in the previous periods.Chapter 2 builds upon chapter 1 to study the nonlinear dynamics of different types of disaggregated capital flows, as well as the effects of capital flow determinants across regimes. Various global and domestic factors are identified as capital flow drivers in the previous literature and are added to the TAR model. Results of this chapter confirmed the nonlinear dynamics of total flows found in chapter 1. Moreover, similar dynamics are found in disaggregated flows and the asymmetry is more pronounced for bank flows. Comparing across countries, the unstable dynamics are present in all flow types for Italy but in fewer flow types for the U.S. and Germany. Regarding capital flow determinants, I find that global factors such as VIX are more significant and have larger effects in crisis times and matter more for bank flows. In contrast, domestic factors are less significant overall and matter more for FDI flows. These findings are also confirmed in principal component analysis. This chapter suggests that unstable dynamics can exist on the disaggregated level causing consequences in specific economic sectors. Also, the effects of capital flow determinants may vary across regimes and flow types. Global factors should be monitored more closely in crisis times especially for the banking sector.Chapter 3 deals with the dynamics of foreign exchange rates. I use a real-time bubble detector to see if there are periods of time in which the actual exchange rate deviates from the fundamental rate in an explosive manner. Different fundamental exchange rates are constructed based on both PPP measures and macroeconomic factors. The results show that bubbles exist in most of the currency pairs in the sample despite being relatively short-lived and infrequent. In addition, I estimate a trivariate VAR model to explore to what extent foreign exchange rates are affected by nonfundamental innovations. The results show that nonfundamental innovations tend to explain a large share of exchange rate variations especially for currency pairs with bubble episodes. The existence of bubbles might reflect speculative activities and interaction of market forces in the foreign exchange market.This dissertation finds nonlinear dynamics exist in both capital flows and exchange rates. The switching to unstable dynamics and the existence of bubbles show the importance of monitoring financial series closely to detect possible shifting of regimes. A future research direction would be to study the reasons for such nonlinear dynamics more in detail and to develop deeper insights to tame excess volatility in the financial market.