Quantitative Finance,2019 Vol. 19, No. 5, 705–716, https: //doi.org/10.1080/14697688.2018.1532101 ©2018iStockphotoLP Economic and political effects on currency clustering dynamics M. KREMER †‡§*††, A. P. BECKER §¶††,I.VODENSKA¶, H. E. STANLEY§ and R. SCHÄFER‡ ∥ †Chair for Energy Trading and Finance, University of Duisburg-Essen, Universitätsstraße 12, 45141 Essen, Germany ‡Faculty of Physics, University of Duisburg-Essen, Lotharstraße 1, 47048 Duisburg, Germany §Center for Polymer Studies and Department of Physics, Boston University, 590 Commonwealth Avenue, Boston, MA 02215, USA ¶Department of Administrative Sciences, Metropolitan College, Boston University, 1010 Commonwealth Avenue, Boston, MA 02215, USA Research and Prototyping, Arago GmbH, Eschersheimer Landstraße 526-532, 60433 Frankfurt am Main, Germany ∥ (Received 20 December 2017; accepted 7 September 2018; published online 13 December 2018) The symbolic performance of a currency describes its position in the FX markets independent of a base currency and allows the study of central bank policy and the assessment of economic and political developments 1. Introduction currency to another currency, using their assets in the mar- ket to accomplish a fixed exchange rate. If a central bank does Similar to other financial markets, exchange rates between not intervene, its currency is considered free-floating, mean- different currencies are determined by the laws of supply ing that the exchange rate is mostly determined by market and demand in the forex market. Additionally, market partic- forces. Some central banks allow their currency to float freely ipants (financial institutions, traders, and investors) consider within a certain range, in a so-called managed float regime. macroeconomic factors such as interest rates and inflation The value of any given currency is expressed with respect to assess the value of a currency. Central banks may par- to the rest of the market through pairwise exchange rates. ticipate in the forex market as well, when pursuing their Currency quotes thus exhibit an important difference to fiscal and monetary policy goals. The degree of central bank equity, fixed income or commodity markets where prices of interventions (CBIs) in the market determines the regime these assets are quoted in one specific currency. A conse- in which a currency trades. Some central banks peg their quence of this is that the appreciation of one currency implies the depreciation of the currency against which it is traded. *Corresponding author. Email: [email protected] This structural property of the market in combination with the ††MK and APB share first authorship. strong influence by macroeconomic fundamentals and central ©2018InformaUKLimited,tradingasTaylor&FrancisGroup 706 Feature banks as market participants have led to specific characteristic In this paper, we present a novel approach to address these behaviors for currencies. issues. Instead of considering currency pairs, we treat each Qualitatively, we can distinguish between hard and soft cur- currency i in a market of K currencies as an individual entity rencies; hard currencies are considered a store of value due to with assigned symbolic performance ζi.Thisapproachintro- their stability even in adverse global economic conditions, and duces a measure independent of base currency to investigate soft currencies are more volatile, for example due to deterio- the hierarchy and the dynamics of the FX market. We aim to rating economic conditions in respective countries. Examples encode the relationship of each currency with the remaining of hard currencies are typically the US dollar, the euro or the currencies in the market into one quantity. Instead of consid- Japanese yen (Hossfeld and MacDonald 2015). The Venezue- ering all currency pairs for a given currency we compress lan bolivar, on the other hand, and its continued devaluation information of its pairwise exchange rates into one quantity over the last decade is an example of a soft currency. for each currency. We do this by measuring the relative per- Alternatively, we can distinguish between reserve curren- formance of a currency in relation to other currencies, and we cies, funding currencies, and commodity currencies. Reserve call this measure the symbolic performance. currencies are currencies which central banks typically hold Using the symbolic performance, we investigate how a cur- as foreign exchange reserves, for which they prefer hard rency’s role evolves within the market in the wake of changing currencies (Habib and Stracca 2012). Most of the world’s economic conditions. As exchange rates are affected by mon- currency reserves are held in the US dollar or euro, and etary policy, we especially consider central banks’ currency to a lesser extent in the British pound and Japanese yen. interventions that may be conducted directly, for example, if As a result, until the recent inclusion of the Chinese yuan, acentralbankpurchasesorsellsthedomesticcurrency.In these four currencies also comprised the currency basket used more extreme cases central banks may introduce a cap on or for accounting purposes at the International Monetary Fund apegofitscurrency,backingthispolicybycurrencytransac- (IMF).† Funding currencies are currencies which can be bor- tions. Data for interventions of this kind are publicly available rowed at low interest rates. Historically, the Swiss franc and for the Swiss franc, the Mexican peso, the Singapore dol- the Japanese yen have been used to fund purchases of cur- lar and Japanese yen among the currencies considered in this rencies with higher interest rates, for example. Commodity paper. currencies are currencies of countries whose economic output In the literature, the effects of CBIs on foreign strongly depends on the price of one or more commodities. exchange rates have been studied by various tech- Examples of commodity currencies include the Norwegian niques, particularly focusing on volatility of exchange krona due to Norway’s significant oil exports or the Australian rates. These techniques include GARCH type models dollar due to Australia’s significant exports of metals and coal. (Almekinders and Eijffinger 1996, Baillie and Oster- It becomes clear that the exchange rate of two currencies berg 1997b,a,Dominguez1998,Beineet al. 2002), implied is therefore determined by their own idiosyncratic behaviors volatility estimation of currency options (Bonser-Neal and and economic factors as well as by their relationships with Tanner 1996,Dominguez1998), regime-switching analysis other currencies in the market. The structure and the charac- of mean and variance of exchange rates (Beine et al. 2003), teristics of the foreign exchange market pose an extraordinary realized volatility estimation (Dominguez 2006,Beineet challenge to traders and researchers. The choice of the trans- al. 2009,Chenget al. 2013), time series study of news reports action or base currency influences the results, which has been (Fatum and Hutchison 2002), and event study of CBIs (Fatum observed in the literature in several works and in many differ- and Hutchison 2002, 2003,Fatum2008). ent contexts. Papell and Theodoridis (2001)analyzeeffortsto Most of the works quoted consider only three currencies— calculate the purchasing power parity (PPP) via panel tests. the German deutschmark (euro), Japanese yen and US They show that the choice of base currency influences the dollar—and study the respective CBIs of the German Bun- outcomes of PPP-tests. Recognizing that the choice of base desbank (European Central Bank), Bank of Japan (BOJ) and currency affects the correlation between different currencies, Federal Reserve System on their domestic currency. Our Hovanov et al. (2004)createacurrencyindextodeterminethe approach, however, explicitly incorporates information of cur- value of an individual currency within the global forex mar- rencies whose central banks did not intervene in the time ket, independent of base currency. Many network approaches period being analyzed. This methodological distinction allows to understanding the foreign exchange markets rely on cor- us to examine not only effects of CBIs on the domestic relation measures. Górski et al. (2008)andKwapien´ et currency, but on the currency embedded in the FX market. al. (2009)investigatetheeffectofbasecurrencyoninterrela- The rest of this paper is structured as follows. We present tionships between currencies when studied through a network our foreign exchange data set in Section 2.Welayoutthe lens. They provide evidence that the topology of the foreign framework and the methodology of obtaining the symbolic exchange network and the structure of its minimum spanning performance measure in Section 3.InSection4,westudythe tree is different for different base currencies. statistics of symbolic performances for the entire time horizon as well as for specific subintervals. In particular, we present the results of our cluster analysis revealing the temporal evo- lution of the symbolic performances and identifying different †The accounting currency of the IMF are the so-called Special roles currencies play within the FX market. We link changes Drawing Rights, and as of 2018 their value is determined through a in roles and behaviors of currencies to central bank interven- weighted basket of the US dollar, euro, Japanese yen, British pound tions as well as economic shocks. We offer our conclusion in and Chinese yuan. Section 5. Feature 707 2. Data We gather information on CBIs from the websites of central banks who