I.1 Motivation Exchange rates are a key issue in international economics and politics. While the determinants of exchange rates have been extensively studied in previous works, this dissertation contributes to the literature by deriving exchange rate expectations from stock market (ADR) data and analyzing their determinants. This exercise is done for three cases where one has to resort to exchange rate expectations since the national exchange rate is either manipulated by the central bank (the first paper in Chapter II), fixed in pegged exchange rate regimes (the second paper in Chapter III), or not existent as the considered country is part of a currency union and therefore has no national currency (the third paper in Chapter IV). The first paper presented in Chapter II analyzes exchange rate expectations for the case of China in the period 1998-2009 in order to test standard exchange rate theories. American officials repeatedly accused China of systematically undervaluing its currency against the U.S. dollar , which produces political tensions between both countries. A recent climax in this dispute was reached on September 28, 2010, when the House of Representatives passed the Currency Reform for Fair Trade Act, which would allow the imposition of import duties for countries with undervalued currencies, namely China. Although this bill did not pass the Senate, Chinese officials clearly opposed the bill arguing against significant undervaluation of the yuan and in favor of political opportunism of U.S. officials. As the assessments of a fair exchange rate significantly differ among officials of both countries, the Chinese-American exchange rate dispute continues. Measuring the development of market determined exchange rate expectations may help to find a compromise in this international political dispute and knowing the determinants of these expectations may help to identify macroeconomic policies necessary to influence future exchange rates. The second paper presented in Chapter III investigates the development of exchange rate expectations and their determinants for the currency crisis episodes in Argentina (2001-2002), Malaysia (1998-1999), and Venezuela (1994-1996 and 2003-2007). Large devaluations of Southeast Asian and Latin American currencies were to be observed during the currency crises in the 1990’s and at the beginning of the last decade. Due to an appreciation of foreign currency denominated debt, capital withdrawals, and bank runs, for example, currency crises typically lead to significant output losses in the affected economies (Hutchison and Noy, 2002). Avoiding currency crisis outbreaks has therefore become one of the major policy goals in many developing countries, which may explain the rapid accumulation of foreign exchange reserves aimed to fend off speculative attacks in these countries. The costs of this currency crisis prevention policy are however often overseen. Since foreign exchange reserves are typically invested in U.S. Treasuries, they yield a relatively low return compared to the high cost of domestic capital in these countries. Moreover, foreign exchange reserves may lose in value as the domestic currency appreciates against the U.S. dollar (Rodrik, 2006). An alternative way to avoid the outbreak of currency crises may be to regularly adjust the official exchange rate (typically managed by the domestic central bank) to levels in line with market expectations. Knowing market-based exchange rate expectations and their determinants may therefore be a cheaper way to avoid currency crises than holding excess amounts of foreign exchange reserves. The third paper presented in Chapter IV uses daily ADR data to analyze the determinants of the risk of withdrawals from the Economic and Monetary Union (EMU) for the five vulnerable member countries Greece, Ireland, Italy, Portugal, and Spain for the period 2007-2009. The subprime lending crisis has triggered significant financial turmoil in the EMU. Banking systems were destabilized and the governments of Greece, Ireland, and Portugal had to be bailed out. Reasserting national authority over monetary policy may help domestic policymakers to address the problems caused by banking and sovereign debt crises or an overvalued euro at national discretion. While the abandonment of fixed exchange rate regimes has so far been analyzed for countries with national currencies, the financial vulnerabilities in the EMU offer a new case to study the possibility of withdrawals from a monetary union. Although a country’s membership in the EMU is typically considered irreversible, many authors agree that sovereign states can choose to leave the EMU (Cohen, 1993; Scott, 1998; Buiter, 1999; Eichengreen, 2007). The new Treaty of Lisbon now includes a provision outlining voluntary withdrawal from the Union, which may cause the members to re-think the pros and cons of remaining in the EMU. Although the European Central Bank (ECB) has implemented measures meant to support the banking sectors and governments in the EMU, autonomous national central banks would probably pursue more expansionary monetary policies. By analyzing the determinants of exchange rate expectations in the monetary union one may therefore analyze the drivers of the risk of withdrawal from the EMU. I.2 Deriving exchange rate expectations from prices of American Depositary Receipts Measuring movements in exchange rate expectations is a relatively easy task for currencies in which a liquid and free forward exchange market exists. For the cases considered in this dissertation, however, the forward exchange market either produces bad forecasts or does not even exist. For the case of China, the yuan/U.S. dollar forward exchange rate is most likely managed by the Chinese central bank in the course of its foreign exchange market intervention policies, which hampers its ability to provide good signals for the future spot market exchange rate (see, for example, Wang, 2010). For the considered member countries of the EMU, no national currencies exist and consequently forward exchange rates cannot be used. For the case of the currency crisis episodes studied in this dissertation, one could use regression-based forecasting models that employ data on macroeconomic variables in order to produce currency crisis signals (see, among others, Eichengreen et al., 1995; Frankel and Rose, 1996; Kaminsky et al., 1998; Kaminsky and Reinhart, 1999; Karmann et al., 2002). The drawback of these approaches is the nature of macroeconomic data used, which enables one to create only monthly or quarterly crisis signals based on backward-looking data. In this dissertation I use stock market data to derive exchange rate expectations, which has several advantages compared to existing approaches. First of all, the prices of the considered stocks are most probably not manipulated by central bank interventions since these stocks are traded in the United States, which enables the derivation of exchange rate expectations formed under free market conditions (also for China). The used stock market data is available for the considered EMU member countries, which facilitates the analysis of the risk of withdrawals from the EMU. Moreover, stock market data is forward-looking and available on a daily basis, which enables the derivation of more accurate and up-to-date currency crisis signals for the considered crisis episodes. In order to derive exchange rate expectations I use data on a particular type of stock called American Depositary Receipt (ADR). An ADR is a financial instrument for foreign companies to list their shares at stock exchanges in the Unites States. An ADR represents the ownership of a specific number of underlying shares of a company in the home market on which the ADR is written. While the underlying stock is denominated in the currency and traded at the stock exchange of the home market, the ADR is denominated in U.S. dollars and traded at a U.S. stock exchange. Since both types of stocks of the same company generate identical cash flows and incorporate equivalent rights and dividend claims, cross-border arbitrage implies that the ADR and its underlying stock have the same price when adjusted for the current exchange rate. When capital controls or ownership restrictions are implemented, cross-border arbitrage is not possible and the law of one price is not binding. In such an environment, information efficiency suggests that the relative prices of ADRs and their underlying stocks – which only differ with respect to the currency they are denominated in – will signal exchange rate expectations of stock market investors. Using data on relative prices (or returns) of ADRs and their underlying stocks and the current exchange rate I can calculate measures for exchange rate expectations of stock market investors. Although the papers presented in this dissertation differ with respect to the considered companies, countries, and time periods, each paper uses the same kind of data and a similar methodology to derive exchange rate expectations – relative prices or returns of ADRs and their corresponding underlying stocks. In each paper I use a panel regression framework in order to analyze the determinants of exchange rate expectations. Each of the included papers focuses on a distinct facet of exchange rate expectations. The first paper focuses on standard exchange rate theories such as the relative purchasing power parity or the uncovered interest rate parity in order to analyze the factors that drive exchange rate expectations in general. The second paper studies the determinants of currency crisis expectations. The third paper analyzes the determinants of the risk of withdrawals from the EMU as expected by ADR market investors. I.3 Contribution to the literature This dissertation adds to two strands of the literature. First, it contributes to a literature that studies the determinants of exchange rates, currency crisis outbreaks, and risk of withdrawal from the EMU. The first paper (Chapter II) contributes to a vast literature on the determinants of exchange rates. An incomplete list of exchange rate determinants analyzed in the literature includes: labor productivity (Chinn, 2000; Cheung et al., 2007); inflation rates (Lothian and Taylor, 1996; Taylor et al., 2001); interest rates (Froot and Thaler, 1990; Chinn, 2006); overvaluation of the domestic currency (Glick and Rose, 1999; Corsetti et al., 2000); or export growth (Williamson, 1994; Isard, 2007). I study the impact of these macroeconomic fundamentals on ADR investors’ exchange rate expectations for China. China makes a good case to study standard exchange rate theories since the Chinese central bank manages the official yuan/U.S. dollar exchange rate, which therefore reacts much less to changes in macroeconomic fundamentals than is suggested by theory. Using ADR market data, I can test exchange rate theories for the Chinese peg/managed float regime under free market conditions. The second paper (Chapter III) contributes to a literature, which analyzes the determinants of currency crisis outbreaks (Eichengreen et al., 1995; Kaminsky and Reinhart, 1999; Karmann et al., 2002). Existing papers employ low-frequent and backward-looking macroeconomic data to forecast currency crises. This dissertation uses ADR market data to derive more accurate and up-to-date currency crisis signals on a daily basis. Moreover, the determinants of currency crisis expectations, such as banking or sovereign debt crisis risk, can be studied using daily market-based risk proxies. The third paper (Chapter IV) contributes to a literature on the sustainability of the EMU. Several papers discuss the possibility of withdrawal from the EMU (Cohen, 1993; Scott, 1998; Buiter, 1999; Eichengreen, 2007). I present empirical evidence that daily ADR market data reflects the risk that vulnerable member countries may leave the EMU and analyzes which determinants drive this withdrawal risk perceived by ADR investors. Second, this dissertation contributes to the literature on the pricing of ADRs. A common finding in the literature is that the outbreak of a currency crisis negatively affects the returns of U.S. dollar-denominated ADRs as the devaluation of the local currency depresses the dollar value of the underlying stock (see, for example, Bailey et al., 2000; Kim et al., 2000; Bin et al., 2004). Several papers find that the introduction of capital controls (typically meant to prevent a currency crisis outbreak) can lead to a permanent violation of the law of one price between ADRs and their underlying stocks since cross-border arbitrage cannot take place (Melvin, 2003; Levy Yeyati et al., 2004, 2009; Auguste et al., 2006). Arquette et al. (2008) and Burdekin and Redfern (2009) find that the price spreads of Chinese cross-listed stocks are significantly driven by market-traded forward exchange rates. This dissertation builds on these findings and uses the relative prices (or returns) of ADRs and their underlying stocks to derive exchange rate expectations. I present empirical evidence that ADR investors’ exchange rate expectations are driven by theory-based determinants of exchange rates, currency crisis outbreaks, or the risk of withdrawal from the EMU. This analysis therefore provides new insights into the price (return) determinants of ADRs. I.4 Main findings and policy implications The findings of this dissertation may broaden the understanding of exchanger rate expectations. The results of the first paper (Chapter II) suggest that stock market investors form their exchange rate expectations in accordance with standard exchange rate theories. Based on a monthly panel data set comprised of 22 ADR/underlying stock pairs and 52 H-share/underlying stock pairs from December 1998 to February 2009 I find that stock market investors expect more yuan appreciation against the U.S. dollar: if the yuan’s overvaluation decreases (the incentive of competitive devaluation); if the inflation differential vis-à-vis the United States falls (relative purchasing power parity); if the productivity growth in China accelerates relative to the United States (the Harrod-Balassa-Samuelson effect); if the Chinese interest rate differential vis-à-vis the United States decreases (uncovered interest rate parity); when Chinese domestic credit relative to GDP decreases (lower risk of a twin banking and currency crisis); or, if Chinese sovereign bond yields fall (lower risk of a twin sovereign debt and currency crisis), ceteris paribus. These findings suggest that the theoretical links between macroeconomic variables and exchange rates in most cases also apply to exchange rate expectations. In this way, the results support the validity of many exchange rate theories and substantiate the rationality of stock market investors’ expectations. This approach (based on stock prices formed under free market conditions) provides an opportunity to test exchange rate theories in managed floating regimes, where the official exchange rate is manipulated by the central bank and does therefore not necessarily respond to changes in macroeconomic fundamentals. Moreover, I use a rolling regressions forecasting framework in order to evaluate the quality of exchange rate expectations. I find that exchange rate expectations drawn from the ADR and H-share market have a better ability to predict changes in the yuan/U.S. dollar exchange rate than the random walk or forward exchange rates, at least at forecast horizons longer than one year. The People’s Bank of China may take advantage of ADR and H-share based exchange rate expectations in order to determine possible misalignments of the yuan/U.S. dollar exchange rate. In this way, the Chinese central bank may improve the timing and intensity of foreign exchange market interventions meant to manipulate the yuan/U.S. dollar exchange rate. The second paper (Chapter III) focuses on the derivation and determination of currency crisis signals formed by ADR market investors. Using daily data on 17 ADR/underlying stock pairs for the capital control episodes in Argentina (2001-2002), Malaysia (1998-1999), and Venezuela (1994-1996 and 2003-2007) we find that ADR investors anticipate currency crises or realignments well before they actually occur. Policymakers could use ADR investors’ up-to-date assessment of the peg’s sustainability in order to identify currency crisis risk earlier and to take the necessary steps to realign an (unsustainable) peg rate before a crisis breaks out. In this way, they could prevent the outbreaks of damaging currency crises without holding excess amounts of costly foreign exchange reserves. Using panel regressions we find that ADR investors anticipate a higher currency crisis risk when export commodity prices fall, the currencies of trading partners depreciate, sovereign bonds yield spreads rise, and interest rate spreads increase. These findings suggest that ADR investors’ currency crisis expectations are based on currency crisis theories even on a daily basis underlining the validity of these theories. The third paper (Chapter IV) studies a particular form of currency crisis risk: the risk that vulnerable member countries could leave the EMU. I use a multifactor pricing model to test whether the financial vulnerability measures assumed to reflect the incentives of national governments to withdraw from the EMU (banking crisis risk, sovereign debt crisis risk, and overvaluation of the euro) are priced in ADR returns. Using daily data on 22 ADR/underlying stock pairs of Greece, Ireland, Italy, Portugal, and Spain in the period January 2007 to March 2009 I find that ADR investors perceive a higher risk of withdrawal (priced in ADR returns) when the risk of banking and sovereign debt crisis and the overvaluation of the euro increase. Policymakers could use ADR market data in order to assess the stability of the EMU. Higher correlations between ADR returns and currency crisis risk factors would suggest a higher risk of withdrawals from the EMU. In such a case, financial vulnerabilities may be addressed within the EMU in order to preserve the integrity of the eurozone. However, time will show how long the policymakers in the EMU will continue with the implementation of even more anti-crisis measures. Growing controversies on the ECB’s sovereign bond purchases and the bailouts for Greece, Ireland and Portugal cast doubt on the sustainability of the EMU in its current form.

Finanzkrisen sind spätestens seit den 1990'er Jahren ein sehr prominentes und interessantes Forschungsobjekt der Wirtschaftswissenschaften. Dieses Jahrzehnt war durch eine Vielzahl von Finanzkrisen gekennzeichnet, von denen nur wenige Länder vollkommen verschont blieben. Die vorliegenden Arbeit gliedert sich in drei Hauptteile: Im ersten Teil werden die unterschiedlichen Theorien zur Beschreibung von Finanzkrisen kategorisiert und jeweils verschiedene bedeutende Arbeiten vorgestellt. Im zweiten Teil der Arbeit werden Gründe für das Auftreten von Herdenverhalten auf Finanzmärkten und dessen Auswirkungen untersucht. Dabei wird zunächst der Stand der Forschung dargestellt. Eine wichtige Voraussetzung für das effiziente Funktionieren von Märkten ist, daß die Marktteilnehmer vollkommene Informationen besitzen. Dies ist in der Realität aber zumeist nicht gegeben. Das in Teil zwei entwickelte Modell basiert auf darauf, daß Informationen unvollkommen sind und asymmetrisch verteilt. Es ist in der Lage die Entwicklung von Herdenverhalten dynamisch zu erklären. Wie dieses Modell zeigt, kann Herdenverhalten zum einen zu stark abweichenden Asset-Preisen führen, welche dabei, gleich einer Blase, stark ansteigen können, um dann wieder ebenso stark zurückzugehen bzw. zu platzen. Zum anderen kann sich ein Herdenverhalten unter internationalen Anlegern auf die Wechselkurse auswirken oder sogar zu einem Zusammenbruch eines fixen Wechselkurses führen. Insbesondere in den 1990'er Jahren konnte man immer wieder beobachten, daß Finanzkrisen nicht nur auf ein einzelnes Land beschränkt waren, sondern sich über mehrere Länder innerhalb einer Region oder auch über die ganze Welt verteilt ausbreiteten. Im dritten Teil der vorliegenden Arbeit werden derartige Ansteckungseffekte genauer untersucht. Hierzu wird zunächst ein Überblick über den Stand der Forschung gegeben. Weiter wird ein Modell entwickelt, welches eine portfoliotheoretische Erklärung für Ansteckungseffekte liefert. Der contagion-Mechanismus kann dabei kurz wie folgt beschrieben werden: Eine Krise in einem Land j führt zu Verlusten eines hoch informierten Investors aus einem Land i. Dies kann wiederum dazu führen, daß er die Investitionen in seinem Heimatmarkt reduziert und dadurch fällt dort der Preis eines Assets. Ein Investor aus einem dritten Land mit einem niedrigeren Informationsstand wird diesen Preisrückgang als schlechtes Signal für den Ertrag aus dem Asset interpretieren. Als Konsequenz zieht er ebenfalls Kapital aus diesem Asset und damit auch aus Land i ab. Der Kapitalabzug des ausländischen Investors kann nun dazu führen, daß es in Land i ebenfalls zu einer Krise kommt. Die Arbeit schließt mit einer Zusammenfassung der gewonnenen Erkenntnisse und liefert einige sich hieraus ergebende Politikempfehlungen zur Verhinderung bzw. Abschwächung von Finanzkrisen. Dabei werden insbesondere die Ursachen und Folgen von Herdenverhalten und Ansteckungseffekten berücksichtigt.