WPS7113 Policy Research Working Paper 7113 What Makes a Currency Procyclical? An Empirical Investigation Tito Cordella Poonam Gupta Development Economics Vice Presidency Operations and Strategy Unit November 2014 Policy Research Working Paper 7113 Abstract This paper looks at the correlation between the cyclical procyclical currencies tend to restrict their capital accounts, components of gross domestic product and the exchange perhaps as an attempt to reduce the degree of procyclicality; rate and classifies countries’ currencies as procyclical if they (iii) countries with procyclical currencies pursue procyclical appreciate in good times, countercyclical if they appreciate monetary policy; (iv) however, in the last decade, there is in bad times, and acyclical otherwise. With this classifi- a disconnect between the cyclicality of currency and mon- cation, the paper shows that: (i) the countries that are etary policy; and (v) the disconnect may reflect a decline commodity exporters and experience procyclical capital in the fear of floating, which can be partially attributed to flows tend to have procyclical currencies; (ii) countries with an improvement in countries’ net foreign asset positions. This paper is a product of the Operations and Strategy Unit, Development Economics Vice Presidency. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at tcordella@worldbank.org and pgupta5@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team What Makes a Currency Procyclical? An Empirical Investigation 1 Tito Cordella and Poonam Gupta Development Economics (DEC), The World Bank, 1818 H St. NW, Washington, DC, 20433 Emails: tcordella@worldbank.org and pgupta5@worldbank.org JEL Classification: F31, E32, F41, E52, E44. Keywords: currency cyclicality, procyclical currencies, countercyclical currencies, exchange rate, monetary policy, trilemma, fear of floating, foreign asset position. 1 We would like to thank participants at the joint IDB-JIMF Conference on “Macroeconomic and Financial Challenges Facing Latin America and the Caribbean After the Crisis,” for vey insightful comments and suggestion. A very special thank goes to our discussant Eduardo Borenztein and to the editor of this special issue, Carlos Vegh, for many detailed suggestions and constructive comments that allowed us to improve the paper significantly. We are grateful to Philip Lane, Donald McGettigan, Guillermo Vulletin and their coauthors for kindly sharing their dataset. We would also like to thank Giovanni dell’Ariccia, Eduardo Levy Yeyati, Andrew Powell, Guillermo Vulletin, Angel Ubide as well as our DECOS colleagues at the World Bank for many useful discussions. We are indebted to Jaime de Jesus Filho and Eleonora Mavroeidi for outstanding research assistance, and to James Trevino for editorial assistance. 1. Introduction If countercyclical monetary policy has been the norm in advanced economies, at least until recently, this was not the case in emerging economies. Why has it been so difficult for emerging markets to adopt countercyclical monetary policies? If institutional weaknesses perhaps constitute one part of the story (Vegh and Vuletin, 2002, and McGettigan et al, 2013), the other common explanations point to specific factors that influence the way emerging markets are affected by, and cope with, the business cycle fluctuations. 2 One such factor that has received the most attention in the literature is the cyclicality of capital flows. Kaminsky, Reinhart and Vegh (2005), KRV henceforth, show that capital flows to emerging markets are procyclical, that is, plentiful in good times and meager in bad times; as a result they tend to pursue procyclical macro policies. The combination of procyclical capital flows and procyclical policies contributes to the “when it rains it pours” phenomenon. 3 While sharing KRV’s view that procyclical capital flows are a key driver of the business cycle in emerging markets, we look at a different variable in this paper: the degree of cyclicality of currencies. More precisely, we compute a currency cyclicality index (CCI) as the correlation between the cyclical components of the nominal (effective) exchange rate and GDP. We show that some currencies, particularly those of emerging markets, have a tendency to appreciate when the business cycle is strong and others, mainly those of advanced economies, have a tendency to appreciate when the business cycle is weak. We christen the former “procyclical,” and the latter “countercyclical.” 4 Even though the degree of cyclicality of a currency differs sharply across emerging and advanced economies, it has not yet received much attention in the literature. The analysis of CCI helps us improve our understanding of the relationship between the cyclical co-movements of economic growth and exchange rate, the correlates of this relationship, and its impact on policies. The relationship between economic activity and exchange rate behavior is a priori ambiguous. In the theoretical literature this relationship is model dependent. In the original Mundell-Fleming framework, if a positive income shock in a country deteriorates the current account, the adjustment occurs through a devaluation of the currency, predicting the correlation between economic growth and exchange rate to be countercyclical (i.e., the currency depreciates in good times). Instead, the correlation is predicted to be procyclical in monetary models, where the value of currency is a function of money demand (Lucas, 1982) and stronger growth increases the (domestic) demand for money under both flexible (Frenkel, 1976) or sticky prices (Dornbusch, 1976). 5 The procyclical 2 Vegh and Vuletin (2012) measure institutional strength by an index based on the international risk guide dataset, while McGettigan et al, 2013 proxy institutional strength by the adoption of inflation targeting and stability of government in the ICRG database. 3 KRV contrast the experience of developing, middle income (loosely overlapping with the set of emerging markets in the parlance that we have adopted in this paper) and developed economies and find that while capital flows are procyclical across all kinds of countries, fiscal policy is procyclical in developing and middle income countries, and the procyclicality of monetary policy is most pronounced in the latter. Thus, the “when it rains it pours” phenomenon, that is, the vicious circle of business cycle and policy cyclicality, affects emerging markets the most. 4 A preliminary discussion of currency cyclicality in emerging markets can be found in Cordella et al. (2014). 5 See Chinn (2013) for a recent survey. 2 behavior of currency is also predicted by the new open economy macro models à la Obstfeld and Rogoff (1995), where an increase in domestic output relative to foreign output leads to a depreciation of the exchange rate—a positive productivity shock increases the volume of the domestically produced varieties and decreases their relative price. The empirical literature does not provide any clear guidance either on the relationship between economic activity and exchange rate behavior. The lack of a clear pattern between the growth rates of consumption and the real exchange rate is documented by Backus and Smith (1993) and subsequently by Chari et al. (1992) and Corsetti et al. (2002). Despite some important progress in empirical exchange rate modeling (see, for instance, Engel et al., 2008), the current knowledge on the behavior of exchange rate and its determinants remains quite limited (Rogoff, 2009). We calculate the CCI using quarterly data for 63 countries over a period of 30 years. Following an identification strategy similar to that in KRV, we assume that the business cycle is driven by the capital flows that the countries receive and by their dependence on commodity exports; we also assume that these exogenous factors affect the CCI. We recognize that the relationship between these factors and the CCI is influenced by the institutional arrangements within the country such as the choice of the exchange rate regime and the degree of openness of its capital account; we discuss how the CCI affects the stance of monetary policy; and how these relationships have evolved over time. We show that the countries that are commodity exporters and experience procyclical capital flows tend to have more procyclical currencies. This implies that currency procyclicality is more likely to be a feature of emerging economies, rather than of advanced countries, since the latter have more diversified exports and possibly experience less procyclical capital flows (due to “fly to quality” considerations). While our results do not indicate any clear welfare implications of a particular degree of cyclicality, they underscore that the degree of currency cyclicality has important implications for the pursuit of macroeconomic policies, especially in emerging economies. Specifically, our results indicate that the countries with procyclical currencies tend to limit the degree of procyclicality by restricting their capital accounts. This partly reflects their fear of floating—that is, their intolerance to excessive variation in exchange rates (Calvo and Reinhart, 2002). The same fear of floating is reflected in a positive correlation between the procyclicality of currency and monetary policy. We observe that the countries with more procyclical currencies tend to pursue more procyclical monetary policy. Our final observation is that in the last decade a disconnect has emerged between the procyclicality of the currency and that of the monetary policy; this can be partially explained by an improvement in the countries’ net foreign asset position, translating into reduced fear of floating. Before proceeding further, an important caveat lector is in order. Even though our analysis provides new interesting perspectives on the interaction between the exchange rate movements and policy challenges, the nature of the available data does not allow us to establish causality. It is thus 3 safer to offer our results as establishing stylized facts, rather than providing causal explanations. We need to leave this difficult task to future research. The rest of the paper is organized as follows. In Section 2, we describe our data and methodology and present the currency cyclicality index that we calculated. In Section 3, we relate the degree of currency cyclicality with the countries’ resource dependence and the nature of capital flows, as well as the conduct of monetary policy. In Section 4 we look at how these relationships have evolved over time and the implications for monetary policy. Section 5 concludes. 2. Data and Measurement of Currency Procyclicality Our data set consists of an unbalanced panel of 63 countries from 1975q1 to 2013q1. The criterion for including countries in the sample is the availability of quarterly real GDP data, required to calculate the short term cyclical co-movements of GDP and the exchange rate. The data availability varies across countries, as reported in Appendix A. For our analysis we compute the currency cyclicality index (CCI), as the coefficient of correlation between the cyclical component of real GDP and the cyclical component of the nominal effective exchange rate (NEER), where the cyclical components are calculated as deviations from the trend computed using an HP filter on log values of the respective series. 6 A positive value of the CCI implies that higher GDP growth is associated with an appreciation of the exchange rate, and lower GDP growth is associated with a depreciation of the exchange rate; a negative value indicates the opposite relationship. 7 We call a currency procyclical if the CCI is positive and significantly different from zero at the 10 percent level; countercyclical if it is negative and significantly different from zero at the 10 percent level; acyclical if it belongs to the residual category of currencies with insignificant CCI. To check the robustness of the significance threshold, we use a 15 percent confidence interval, as well as a one tailed test, and find the classifications according to the different significance criteria to be very similar (Appendix Table C.1). We also calculate the CCI using the real effective exchange rate and find a large overlap with the one calculated using the nominal effective exchange rate. 8 The CCI calculated using NEER is presented in Figure 1a, and the one using REER is in Figure 1b. 6 We run the HP filter (with lambda equal to 1600) after seasonally adjusting the series using Census X11. 7 Since nominal exchange rates are relative prices we have a numéraire problem and we lose a degree of freedom, which in principle could affect the interpretation of the results. However the problem is not very severe in our analysis, which is based on nominal effective exchange rate with weights that differ across countries. We would like to thank Andrew Powell for raising this important point. 8 The Spearman rank correlation between NEER and REER procyclicality measures is 0.86. 4 From a simple inspection of Figures 1a and 1b, it is evident that the sets of countries that are procyclical, countercyclical and acyclical according to the real or nominal exchange rate based indices overlap significantly, as one would indeed expect in (high frequency) quarterly data. To get a clearer image of the phenomenon that we are discussing here, we plot the cyclical components of GDP and NEER for Brazil, a country with procyclical currency, and Japan, a country with countercyclical currency, in Figure 2a and 2b below. We find preponderance of procyclical currencies among emerging economies and of countercyclical currencies among advanced economies. While many of the advanced economies such as Switzerland, Germany/Eurozone, Japan and the US (in the more recent period) have countercyclical currencies, we find no emerging economy with independent countercyclical currencies. The only emerging economies with a countercyclical currency in our sample are Euro Zone periphery countries that “inherited” a countercyclical currency—the euro has indeed the same degree of cyclicality as the Deutsche Mark. Despite the preponderance of emerging markets in the procyclical currencies group, the group is a bit more mixed. While virtually all the emerging market commodity exporters such as Chile, Argentina, Brazil and Russia are in the group of procyclical currencies, we also find that some of the advanced commodity exporting countries, such as New Zealand, Australia, Finland and Iceland, have procyclical currencies. 9 9 An interesting counter example is Norway, which, despite being one of the larger commodity exporters in the world, relative to the size of its economy, maintains a countercyclical currency. This is seemingly because of the operations of the Norway oil fund; see, e.g., Velculescu (2008). 5 CCI CCI 0.0 0.2 0.4 0.6 0.8 1.0 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.6 -0.4 -0.2 Japan Switzerland Estonia Lithuania Switzerland Netherlands Ireland France Netherlands Costa Rica Lithuania Cyprus Latvia Estonia China HK Norway Bulgaria Germany Luxembourg Belize Japan Latvia Germany Ireland Cyprus Luxembourg France Austria United States Belgium Belize United States Austria Portugal Costa Rica Malta Bolivia Italy Macedonia Philippines Belgium Croatia China Mainland Slovenia Norway Morocco Morocco Canada Canada Georgia Italy Bolivia Malta China Mainland Slovenia Israel Procyclical United Kingdom 6 Croatia Procyclical Peru Denmark South Africa Georgia Denmark South Africa Hungary United Kingdom Australia Russia Poland Spain Greece Poland Acyclical Iran countercyclical (negative and significant correlation) and acyclical (non-significant). Hungary Acyclical Romania Australia India India Israel Peru Bulgaria Greece Macedonia Portugal Russia Slovakia Finland Colombia Spain Chile Colombia Brazil Brazil Czech Repub Iceland Finland Countercyclical Sweden Iran Countercyclical New Zealand Romania Thailand Sweden Malaysia Argentina Turkey Iceland Ukraine Philippines Indonesia New Zealand Mexico Mexico Chile Thailand Argentina Malaysia Uruguay Turkey Figure 1b: Currency Cyclicality Index (CCI), for Real Effective Exchange Rate Korea Rep Ukraine Slovakia Uruguay Figure 1a: Currency Cyclicality Index (CCI), for Nominal Effective Exchange Rate Czech Repub Indonesia Korea Rep Note: Correlations between exchange rates and GDP are calculated using quarterly data and HP trend filtered series. Two-tail tests with significance at 10 percent level are used to divide the countries into procyclical (positive and significant correlation), Figure 2a: Cyclical Comovement of GDP and Exchange Rate– Brazil Brazil: NEER and GDP cycl. comp. .04 .6 .4 .02 NEER .2 GDP 0 0 -.02 -.2 -.04 1995q1 2000q1 2005q1 2010q1 2015q1 Quarters GDP NEER Figure 2b Cyclical Comovement of GDP and Exchange Rate – Japan Japan: NEER and GDP cycl. comp. .04 .2 .02 .1 0 NEER GDP 0 -.02 -.1 -.04 -.06 -.2 1980q1 1990q1 2000q1 2010q1 Quarters GDP NEER Note: Left y-axis measures the cyclical component of quarterly real GDP; right y-axis measures the cyclical component of the nominal effective exchange rate (NEER). HP filter is used to extract the cyclical components. 7 3. Which Country Characteristics Are Correlated with the Currency Cyclicality Index? In what follows, we exmaine certain country characteristics that tend to be associated with procyclical, countercyclical or acyclical currencies. As is often the case in empirical macro analysis, it is difficult to argue convincingly about the exogeneity of any particular variable that we consider below. This, however, does not prohibit us from assuming that certain variables such as the cyclicality of capital flows and the intensity of commodity exports are “more exogenous” than other variables and can be thought of as the “determinants” of the CCI. 3.1 The “Determinants” of the CCI Below, we examine the relationship between the cyclicality of capital flows and the CCI. As is customary in the literature, we measure the degree of cyclicality of capital flows by the correlation between the cyclical component of the quarterly GDP and that of the private net capital inflows. 10 Just as observed in Bluedorn et al. (2013) emerging markets seem to experience more procyclical capital flows. In our data set, the average value for the index of capital flows cyclicality is 0.33 for emerging markets and 0.03 for advanced economies. We observe a strong positive correlation between the procyclicality of capital flows and the procyclicality of the currency, see Figure 3. Figure 3: Cyclicality of Currency and the Cyclicality of Capital Flows 1 CCI = 0.05 + 0.41*Capital Flow Cyclicality (1.35) (2.91) KOR IDN URY UKR TUR MYS .5 THA MEX PHL NZL ARG ISL SWE FIN CZE CHL SVK BRA CCI COL PRT PERGRC AUS IND HUN ESP GBR DNK HRV ZAF ISR 0 CAN MLT ITA MAR NOR BEL AUT CRI FRA CYP USA DEU LUX JPN HKG BGR LVA NLD IRLCHE EST -.5 -.4 -.2 0 .2 .4 .6 Capital Flow Cyclicality Note: CCI is the correlation between the HP filtered NEER and HP filtered real GDP. Procyclicality of capital flows is measured by the correlation between the cyclical components of the quarterly nominal GDP and the cyclical component of the nominal total private net capital inflow. The t-statistics are reported in parentheses. 10The data for capital flows come from Bluedorn et al. (2013). The database includes data on quarterly gross inflows, net capital inflows and GDP in current US$, from 1970 Q1 to 2012 Q2. As before, the cyclical components are calculated as deviations from the HP filter, after both series have been seasonally adjusted. 8 Inspired by the large literature on commodity currencies, next we relate the cyclicality of a country’s exchange rate to its exposure to shocks to commodity prices. 11 A priori, one would expect large commodity exporters to have procyclical currencies. The intuition is that they tend to grow faster when the demand for commodities and their prices are stronger, and the currency tends to appreciate because of the favorable movement in the terms of trade. To test whether this is indeed the case, we construct an index of the “intensity” of commodity exports following Lederman and Maloney (2008). The index, constructed as the net exports of commodities, as percent of GDP, averaged over the sample period, takes positive values if a country is a net commodity exporter and negative values if it is a net commodity importer. As expected we find a positive and significant correlation between the commodity export intensity index and the CCI (Figure 4). This positive correlation seems to transcend across the level of development of the countries in our sample. As noted above we find several advanced countries, which are commodity exporters, to have procyclical currencies. Figure 4: Currency Cyclicality Index and Commodity Exports Index 1 CCI = 0.10 + 10.25*Commodity Export (2.90) (2.29) KOR IDN URY UKR TUR MYS .5 THA MEX PHL NZL ARG ISL ROM SWE FIN IRN CZE BRA CCI SVK COL CHL PRTGRC PER INDHUN AUS ESPPOL RUS GBR GEO HRV DNKZAF ISR SVN 0 MLT ITA MAR CAN CHN NOR MKD BEL AUT CRI BOL USA BLZ CYP LUX JPN DEUFRA BGR LVA LTU NLD CHE IRL EST -.5 -.02 -.01 0 .01 .02 Commodity Exports Index Note: CCI is the correlation between the HP filtered NEER and HP filtered real GDP. The commodity exports index is constructed as the net exports of commodities, as percent of GDP, averaged over the sample period. The t-statistics are reported in parentheses. The fact that the countries with procyclical currencies experience procyclical capital flows, and have a large commodity base, suggests that currency procyclicality is more likely to be a characteristic of developing or emerging economies rather than of advanced countries. In the latter case, the production base is generally more diversified and, more importantly, because of fly to quality considerations, their capital flows tend to be less procyclical and may even be countercyclical. Figure 5 shows that currency procyclicality is indeed a feature of emerging economies. The figure 11 See, among others, Chen and Rogoff (2003) and Cashin et al. (2004). 9 depicts a negative and significant correlation between the CCI and per capita GDP. An increase of 1 standard deviation (1.23) in log GDP per capita is associated with a decline in the CCI of about 0.10. 12 Figure 5: Cyclicality of Currency and GDP Per Capita 1 CCI = 0.84 - 0.08*Log(GDP per capita) (3.15) (2.89) IDN KOR URY UKR TUR MYS .5 THA MEX PHL NZL ARG ISL ROM SWE IRN FIN BRA CZE CCI COL CHL SVK PER PRT GRC IND AUS RUS POLHUN ESP GBR GEO ZAF HRV DNK ISR SVN 0 MAR MLT ITA CAN CHN NOR BOL MKD CRI BEL AUT BLZ USA CYP FRA DEU JPN LUX BGR HKG LVA LTU NLD IRL CHE EST -.5 6 7 8 9 10 11 Log GDP per Capita Note: CCI is the correlation between the HP filtered NEER and HP filtered real GDP. GDP per capita is at US$ 2005 prices. The t- statistics are reported in parentheses. 3.2 CCI and the Policy Stance Next, we look at the way the degree of cyclicality of a currency relates to the choice of the broad policy framework, such as the exchange rate regime, and the degree of openness of the capital account. On the relationship between currency cyclicality and exchange rate regimes, one would expect that, by adopting a peg, a country reduces the volatility of exchange rate and thus the degree of currency cyclicality. This is what we observe in the data as well. Using Reinhart and Rogoff’s “coarse” classification of de facto exchange rate regime (which takes values from 1 for “strongly fixed” to 6 for “strongly flexible” regimes), we find a positive relationship between the flexibility of the exchange rate regime and the degree of procyclicality of the currency, 13 as depicted in Figure 6. In addition, we find that the exchange rate regime conditions the impact of capital flows on the cyclicality of currency. Countries with flexible exchange rate regime (e.g. those above the median) have a positive and significant correlation between the cyclicality of capital flows and that of 12 On the other hand, we do not find a robust correlation between the CCI and the economic size of a country, measured as a country’s share of world GDP. 13 More precisely, one should expect a positive relation between the flexibility of the exchange rate regime and the absolute value of the degree of cyclicality of a currency. The result in figure 6 is driven by the fact that countries with procyclical currencies outnumber those with countercyclical ones in our sample. 10 the currency, and those with less flexible exchange rate regime (e.g. those below the median) have an insignificant correlation between the cyclicality of capital flows and that of the currency. 14 Figure 6: Currency Cyclicality Index and Exchange Rate Regime 1 CCI = -0.14 + 0.10*Exchange Rate Regime (1.47) (2.80) KOR IDN URY UKR TUR MYS .5 THA MEX PHL NZL ISL ARG SWE ROM FIN IRN CZE BRA CCI SVK COLCHL GRCPRT PER IND HUN AUS ESP POL RUS GBR DNK GEO ZAF SVN HRV ISR 0 ITA MARCHN CAN MLT NOR BLZAUT CRI MKD BOL USA FRA CYP DEU JPN HKG BGR LVA NLD LTU IRL CHE EST -.5 1 2 3 4 5 Exchange Rate Regime Note: Exchange rate regime takes values from 1 to 6, 1 indicating “strongly fixed” and 6 “strongly flexible” exchange rates. The t- statistics are reported in parentheses. The implications that follow are that countries can influence the degree of cyclicality of their currency by choosing their exchange rate regime, as well as the degree of openness of their capital account. This is corroborated in Figure 7, which shows that the countries with more procyclical currencies seem more inclined to impose controls on the capital account (we use the Chinn and Ito’s index of capital account openness). 15 Using a common parlance, these results indicate that the countries with procyclical currencies tend to “lean against the wind” by maintaining controls on their capital account. 14 Results are available upon request. 15 The Chinn and Ito index is based on binary dummy variables that codify the tabulation of restrictions on cross-border financial transactions reported in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions. The last update of the index (2011) gives a score of 2.44 to the “most financially open” country, and a score of -1.86 to the “least financial open” country. The annual data for this index goes from 1975 to 2011 and is obtained from http://web.pdx.edu/~ito/Chinn-Ito_website.htm. 11 Figure 7: Currency Cyclicality Index and Openness of Capital Account 3 Capital Openness = 1.04 - 1.57*CCI (6.96) (3.17) IRL NLD CHE EST USA CAN DEU URY LTULVA JPN BEL GRC PER NZL GBR 2 Capital Account Openness AUT FINSWE DNK CZE FRA AUS CRI ITA HUN ESP NORSVN IDN ROM 1 BOL GEO PRT BGR HRV MYS CYP ISL MEX MLT ISR ARG MKD 0 SVK KOR RUS POL CHL THA PHL BRA COL TUR -1 MAR IRN CHN ZAF IND BLZ UKR -.5 0 .5 1 CCI Note: CCI is the correlation between the HP filtered NEER and HP filtered real GDP. The t-statistics are reported in parentheses. One of the key observations we make is that the cyclicality of currency and that of monetary policy are positively correlated—the countries with procyclical currencies tend to pursue procyclical monetary policies, and countries with countercyclical currencies tend to pursue countercyclical monetary policies, Figure 8. While the level of development of an economy is likely to have some bearing on it, the relationships observed between currency cyclicality and policy variables point at the trilemma kind of considerations–the constraint that countries face in the policy space of capital account openness, exchange rate regime and monetary policy. 16 It seems that the emerging markets that are more prone to procyclical currencies resolve the trilemma by choosing limited exchange rate flexibility, a less open capital account, as well as a more procyclical monetary policy. 16 See, among others, Obstfeld et al. (2005), Aizenman et al. (2013), and Klein and Shambaugh (2013). 12 Figure 8: Currency Cyclicality Index and Monetary Policy Procyclicality URY Monetary Policy Cyclicality=-0.28+0.35*CCI .4 (6.25) (2.25) CRI MEX .2 Monetary Policy Proyclicality IND PER CHL BGR ISR PRT TUR GRC 0 THA MAR CHN CYP PHL IRL -.2 BRA ZAF FIN KOR ESP FRA NOR JPNAUT CZE NZL -.4 GBR SWE DNK COL AUS MYS ITA -.6 CHE DEUBEL CAN USA NLD -.5 0 .5 1 CCI Note: CCI is the correlation between the HP filtered NEER and HP filtered real GDP. The cyclicality of monetary policy is calculated as the correlation between the cyclical components of real GDP and short-term interest rates. The t-statistics are reported in parentheses. It would be apt to add here that, even though our analysis primarily looks at correlations, the positive correlation between currency and monetary policy procyclicality seems to have a “causal flavor.” The typical reverse causality argument that the monetary policy stance affects the cyclical behavior of a currency would indicate a relationship quite opposite from the one we are suggesting. A procyclical monetary policy would reduce the procyclicality of capital flows (by increasing their remuneration in bad times and decreasing it in good ones) and thus reduce the degree of procyclicality of a currency. This implies that, if anything, our estimated positive correlation between currency and monetary policy cyclicality is likely to be biased downward. Finally, even though there are no clear a priori theoretical reasons to expect a link between the currency and fiscal policy cyclicality, we examine if the two are correlated in the real world. We compute the cyclicality of fiscal policy as the correlation between the cyclical components of real GDP and real total central government expenditure, as in Vegh and Vuletin (2012) and, perhaps unsurprisingly, we do not find it to be correlated significantly with the degree of cyclicality of the currency (not reported). 4 Currency Cyclicality over Time Moving from the cross sectional dimension of the correlates of the cyclicality of currencies and their implications for economic policy, in this section, we focus also on the time series dimension of the CCI. In particular, we are interested in knowing whether currency cyclicality has evolved over time; whether its relationship with the policy variables and other correlates has changed over time; and how do the changes we observe reconcile with the monetary policy graduation phenomenon observed in Vegh and Vuletin (2012). 13 4.1 Is CCI Persistent over Time? In order to look at the persistence of currency cyclicality, we split our sample in to two equal sub-periods, 1985-1999 and 2000-2013, and we compute the CCI in each of them. The scatterplots for the CCI in Figure 9 indicate a significant degree of persistence. Most of the countries that were countercyclical in the first period remained countercyclical in the second, while only a few moved from being countercyclical in the first period to procyclical in the second, one such notable exception being India. While some countries that were procyclical in the first period became countercyclical in the latter period, most of these are countries that changed their status after joining the Eurozone. 17 Figure 9: Persistence of Currency Cyclicality Index Over Time -1 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 1 CCI = -0.02 + 0.33*CCI(-1) (0.40) (2.91) GBR SWE NZL ARG CHL TUR KOR CAN ISR ROM BRA MEX IND ISL COL AUS CCI: 2000-2013 POL HRV RUS PHL CZE THA SVK HUN NOR IDN SVN IRN MLT PER MYS MARZAF GEO LVA BOL ESP IRL AUT DEU BGRCYP HKG LUX FIN ESTJPN NLD FRA ITA CHE DNKUSA LTU BEL PRT -1 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 1 CCI: 1985-1999 Countercyclical both periods Procyclical both periods Acyclical both periods Other Fitted values Note: CCI is the correlation between the HP filtered NEER and HP filtered real GDP. The black dashed line is the 450 line. The t- statistics are reported in parentheses. 17Interestingly, most of the Eurozone periphery countries (including those that experienced financial difficulties in the recent past such as Italy, Spain, Portugal, and Cyprus) are countries that had a procyclical currency in the first period but, with the adoption of the euro, they inherited the countercyclicality of the old German Mark, which did not change with the euro adoption. For what it is worth, also notice that, while the US and the UK were both acyclical in the first period, they became, respectively, countercyclical and procyclical in the second period. 14 4.2 CCI and Its Correlates over Time We now explore the time variation in the relationship between the CCI and its correlates, comparing the pre and post 2000 periods. The results in Figure 10 suggest that while the relationship between the CCI and exogenous variables—that is, the cyclicality of capital flows and the commodity export base—is stable over time, the relationship between the CCI and the policy variables has changed somewhat. The results particularly indicate that the countries with more procyclical currencies have become more prone to maintaining flexible exchange rate regimes and open capital account in the latter period. This is suggestive of a decline in the fear of floating. Figure 10: Currency Cyclicality Index and its Correlates over Time A. CCI and Commodity Exports B. CCI and Procyclicality of Capital Flows 1 1 GBR GBR KOR IDN KOR IDN SWE SWE NZL NZL ARG KORCHL URY TUR KOR TUR URY CHL UKR TUR UKR TUR ARG MYS THA .5 CAN ISR MEX MEX MYS PHL ISL ISL NZL IND BRA .5 ISR THAMEX MEX CAN ARG ROM PHL IND BRA NZL ISL ISL SWE AUS COL COL ARG AUS FIN CZE HRV PHL ROM SWE CZE CHL BRA COL PHL HRV CZE FIN IRNRUS THA SVK CCI CZE POL BRA CHL PRT PERGRC IND SVK THA COL ESPAUS SVK HUN HUN CCI PRTGRC PER NOR IDN SVK IND ESPHUN HUNPOL AUS GBR IDN RUS NOR DNK HRV ISR ZAF GBR 0 MLT GEO ISR HRV DNKZAF IRN CAN NOR ITA MAR SVN BEL MLT AUT MYS 0 MLT SVN ITA MARCHN CAN NOR FRA CRI USA PER CRI CYP DEU LUX JPN ZAF MLT MKD BEL AUTUSA CRI BLZ PER MYS BOL HKG ESP MAR LVA LVA BGR CYP LUX GEO JPNDEU FRA ZAF CYP AUT IRL HKG NLD BGR MAR LVABGR BOL DEU IRLCHE FIN LUX CYP ESPLVA FRA ITA JPN NLD EST EST LTUAUT IRL BGR FIN NLD CHE -.5 LUX JPN DEU CHE IRL BEL DNK USA ITA FRA ESTEST NLD CHE PRT -.5 BEL LTUUSA DNK PRT -.5 0 .5 1 Capital Flow Cyclicality -.01 0 .01 .02 Commodity Export Full Period Full: CCI = 0.05(1.35)+0.41(2.91)*Capital Flow Cycl. Sub Period: 2000-2013 Sub: CCI = -0.02(0.36)+0.40(2.12)*Capital Flow Cycl. Full Period Full: CCI = 0.10(2.90)+10.25(2.29)*Commodity Export t-stats in parenthesis Sub Period: 2000-2013 Sub: CCI = 0.03(0.66)+10.72(2.04)*Commodity Export t-stats in parenthesis C. CCI and Capital Controls D. CCI and Exchange Rate Regimes 3 DNK PRT USA NLD CHE IRL ITA FRA NLD CHE JPN AUT ESTFIN DEUIRL ESP PER CAN DEU USA NOR GRC CAN URYNZL SWE GBR BEL EST LTU LVA 1 LVA JPN BEL PER NZL LTU Capital Account Openness GBRHUN CZE ISR 2 AUT GBR SVN FIN SWE CHL KOR IDN DNK CZE SWE NZL FRA AUS URY CHL KOR URY TUR CRI ITA HUN ESP ROM UKR ARG UKR TUR BOL SVN MYS .5 CYP NOR AUS IDN THA ISR CAN MEXMEX GEOBOL IDN ROM MEX PHL NZL BRA ISL ROM 1 MLT GEO PRT HRV IND ARGISL BGR SWE COL AUS ROM BGR MYS FIN HRV RUS PHL IRN CZE CZE CHL CYP HRV ISL SVK COL POL THA BRA CCI MLT MEX ISL GRC GRCPRT IND PER AUS ISR SVK ARG ESP SVK HUNHUN POL RUS MKD KOR GBRNOR IDN ISR GEO ZAF 0 SVK BRA IRN DNK SVN HRV MLT PHL POL 0 MYS RUS KOR SVN ITA CHN MARCHN CAN NOR IRN RUS CHL COL POLTHA PHLTHA MYS BRA ARG BLZAUT CRI MLTPER MKD MKD BOL CRI USA FRACYP GEO DEU JPNZAF COL TUR MAR BGR LVA TUR ESP HKG CYP BOL LVA IRN IRL NLD BGR AUT LTU -1 MAR HKG FIN DEU IRL CHE MAR ZAF CHN ZAF IND IND ITA FRA EST NLD EST JPN BLZ UKR CHE -.5 DNK LTU USA PRT -.5 0 .5 1 CCI 1 2 3 4 5 Exchange Rate Regime Full Period Full: Capital Openness = 1.04(6.69)-1.57(3.17)*CCI Full Period Full: CCI = -0.14(1.47)+0.10(2.80)*Exchange Rate Reg. Sub Period: 2000-2013 Sub: Capital Openness = 1.22(7.79)-1.06(2.62)*CCI Sub Period: 2000-2013 Sub: CCI = -0.42(3.64)+0.22(3.61)*Exchange Rate Reg. t-stats in parenthesis t-stats in parenthesis These patterns confirm the observation made in Vegh and Vuletin (2012) that even if a large number of developing countries have continued to pursue procyclical monetary policies, several of them have “graduated” in the last decade and have adopted a countercyclical monetary policy stance. They attribute this transition to a decline in their fear of floating but also indicate that many emerging markets that have transitioned have done so after adopting an explicit inflation targeting 15 framework. 18 This, together with the fact that the transition has occurred while maintaining open capital accounts, reflects an improvement in the quality of their institutions; this strength is further reflected in the waning of the fear of exchange rate volatility. 19 We notice a related phenomenon below. We observe that a “disconnect” has occurred between the procyclicality of currency and monetary policy since 2000. In Figure 11, Panel A, while the CCI and monetary policy procyclicality are positively correlated when calculated for the entire period or for the pre-2000 period separately, the correlation is insignificant in the post-2000 period. The same phenomenon is observed when we use rolling observations. Specifically, we calculate the CCI over overlapping five year periods, each time dropping the data for the first four quarters and adding the data for four quarters at the end. We similarly calculate the rolling averages of the monetary policy procyclicality. 20 The same disconnect that we observed earlier between the procyclicality of monetary policy and the CCI is seen in the rolling observations as well. In Figure 11, Panel B, below, the linear fit between monetary policy procyclicality and the CCI is positive and significant when estimated for the entire period, but insignificant when estimated for the data since 2000. Figure 11: Monetary Policy Cyclicality and Currency Cyclicality Index over Time B. Monetary Policy Cyclicality and CCI – A. Monetary Policy Cyclicality and CCI Rolling data URY .5 .5 URY CRI Monetary Policy Cyclicality CRI IND Monetary Policy Cyclicality MEX ISR MAR IND CHN PERCHL BGR ISR PRT AUS TUR GRC TUR 0 0 THA MAR CHN BRA IRL CYP PHL ZAF BRA FIN KOR FRA ESPTHA PHL CYP JPNAUTNOR CZE NZL DNKGBR SWEMEX JPN BGR AUS COL -.5 PER NOR -.5 MYS MYS KOR CHE CHE DEU ZAF ITA BEL CAN USA COL CAN NZL NLD SWE USA CZE GBR CHL DNK -1 -1 -.5 0 .5 1 -1 -.5 0 .5 1 CCI CCI Full Period Full: Monetary Pol. Cycl. = -0.28(6.25)+0.35(2.25)*CCI Pre 2000 Post 2000 Sub Period:2000-2013 Sub: Monetary Pol. Cycl. = -0.36(5.12)+0.13(0.73)*CCI All: MPC= -0.38(23.11)+0.18(5.38)*CCI Post 2000: MPC = -0.36(10.50)+0.04(0.60)*CCI t-stats in parenthesis. CCI = Currency Cyclicality Index t-stats in parenthesis Similar trends are observed in Table 1 below where, using the rolling observations, we regress monetary policy cyclicality on the CCI interacted with a dummy for the pre 2000 period, and 18 Vegh and Vuletin (2002) measure the fear of floating as the correlation between the cyclical component of the short- term interest rate and the rate of depreciation of the exchange rate. 19 A similar point is made in Adler and Magud (2013) who show that better policies, particularly the sizable increase in aggregate saving rates, may explain why Latin America did relatively well during the recent commodity boom and may also be able to face better the downturn. 20 Rolling regressions provide more robust estimates of how the CCI has evolved and its relationship with other correlates changed over time. 16 on the CCI interacted with another dummy for the post 2000 period. We estimate these regressions including either time fixed effects or a dummy for post 2000, with and without country fixed effects. Results quite decisively show that while the CCI and monetary policy procyclicality are positively correlated prior to 2000, their correlation is insignificant in the post 2000 period. Table 1: Monetary Policy and Currency Cyclicality Index over Time (1) (2) (3) (4) Dependent Variable Monetary Monetary Monetary Monetary Policy Policy Policy Policy Cyclicality Cyclicality Cyclicality Cyclicality CCI x Dummy pre 2000 0.25*** 0.25*** 0.20** 0.22*** (3.24) (3.14) (2.70) (3.03) CCI x Dummy post 2000 0.09 0.04 0.01 0.07 (0.79) (0.31) (0.07) (0.58) Dummy post 2000 0.02 -0.08 (0.28) (1.13) Constant -0.38*** -0.38*** -0.35*** -0.31*** (7.02) (9.24) (18.92) (4.84) Observations 910 910 910 910 Time Effect Yes No No Yes Fixed Effect No No Yes Yes R-squared 0.05 0.04 0.03 0.05 Number of Countries 40 40 40 40 Robust t statistics (clustered by country) are in parentheses. ***, **, * indicate significance at 1, 5 and 10 percent levels respectively. Since the rolling observations introduce serial correlation, we cluster our standard errors by countries in the estimations using these data In an endeavor to understand the factors behind the trends observed in the conduct of monetary policy and its relationship with the cyclical currency movements we make three more observations. First, the countries with procyclical currencies and negative net foreign currency positions (using the Lane and Shambaugh’s, 2010, measure of net foreign currency position) are the one that pursue procyclical monetary policy. 21 This is evident in Chart 12, Panel A and B, where we find a positive and significant relationship between the net foreign asset position and the cyclicality of monetary policy for countries with a positive CCI but no distinct relationship for countries with a negative CCI. This is consistent with the hypothesis that the countries with procyclical currency and currency mismatches in their balance sheet fear large movements in the exchange rate and use monetary policy to tame them. This may imply tightening their monetary policy in the midst of an economic 21Lane and Shambaugh (2010) calculate net foreign currency position as the weighted sum of net foreign assets to capture the sensitivity of a country’s external balance sheet to a uniform movement of its domestic currency against all foreign currencies. The variable takes values between -1 and 1, where a value of -1 corresponds to a country that has zero foreign-currency foreign assets and it only holds foreign-currency foreign liabilities, a country afflicted by an extreme form of original sin. A value 1 corresponds to a country that has foreign assets only in foreign-currency and foreign liabilities only in domestic-currency. The figures we use are those updated by Benetrix et al. (2014). 17 slowdown to avoid depreciation, or loosening their monetary policy in the midst of an economic boom to avoid excessive appreciation. Our second observation, consistent with Benetrix et al. (2014), is that the countries have increasingly improved their net foreign currency positions over time. Improving net foreign currency position depends crucially on the relation between currency cyclicality and the debt dynamics of a country, and thus it may imply different things for different countries. If a country with a procyclical currency issues debt in foreign currency, the cost of servicing debt would be low in good times (when the domestic currency is strong) and high in bad times (when the domestic currency is weak). This means that, for the debt service or consumption smoothing argument, a country with a procyclical currency would be better off if it has a long net foreign currency position. For the same reason, countries with countercyclical currencies should be better off if they have a short net foreign currency position. 22 While in the 1990s, consistent with the original sin literature (Eichengreen et al. 2007), countries with procyclical currencies, or the emerging economies, tended to be very long in domestic currency, the situation has changed dramatically in the last decade. Nowadays, countries with procyclical currencies have a longer net foreign asset position. Panel C in Figure 12 shows this marked increase in the average net foreign currency positions of countries with a positive CCI, while countries with a negative CCI didn’t change much their net foreign asset position or moved to a slightly shorter one. Lane and Shambaugh (2010) note that the transition of emerging markets to more positive foreign currency position has been made possible partly by their improved current account positions and larger foreign reserves, a shift in capital flows to equity from debt, as well as by the success in developing local currency debt markets. Adoption of an inflation targeting framework by countries could be another reason for this transition, as noted in McGettigan et al (2013). Evidence therein shows that the countries that have successfully transitioned to countercyclical monetary policy are the ones that have adopted the inflation targeting framework, possibly reflecting strengthened monetary institutions allowing them to pursue independent monetary policy. We obtain similar regression results which show that the countries that have adopted an inflation targeting framework have less procyclical monetary policy. 22 Lane and Shambaugh (2010) indeed pointed out that, for countries whose exchange rates depreciate after a negative shock, there is an advantage in holding a net long position in foreign currency, since the latter acts as a hedge against asymmetric shocks. 18 Figure 12: Foreign Currency Position and Monetary Policy Procyclicality A. Foreign Currency Position and B. Foreign Currency Position and Monetary Monetary Policy cyclicality. CCI >0 Policy cyclicality. CCI<0 .5 .2 MPC = -0.15-0.79*FXAGG URY MPC=-0.47-0.08*FXAGG (2.99) (2.85) (6.19) (0.21) Monetary Policy Cyclicality Monetary Policy Cyclicality 0 MEX IND PER CHL PRT ISR TUR GRC IRL -.2 0 THA FRA JPN PHL AUT NOR BRA -.4 ZAF FIN KOR ESP CZE NZL SWE GBR COL DNK ITA -.6 AUS BEL DEU CHE CAN USA -.5 MYS NLD -.4 -.2 0 .2 .4 -.2 0 .2 .4 Foreign Currency Position Foreign Currency Position C. Foreign Currency Position Median Evolution .2 Foreign Currency Position - Median -.1 0 -.2 .1 1990 1995 2000 2005 2010 year Positive CCI Negative CCI Advanced Countries Emerging Countries Note: The t-statistics are reported in parentheses. 5 Conclusions International finance literature is quite inconclusive as to whether currencies are expected to appreciate or depreciate over the business cycle. While theoretical contributions offer different predictions, which are often model dependent, no clear pattern has yet emerged from the empirical literature. In an attempt to fill this gap, in this paper we develop an index of currency cyclicality, and use it to study the cyclical behavior of a large number of currencies, over many decades. We unveil a number of empirical regularities. First, countries with procyclical currencies—the currencies that have a penchant to appreciate in good times and depreciate in bad ones—tend to be 19 more exposed to commodity price shocks and are more likely to receive procyclical capital inflows than countries with countercyclical currencies. This very simple characterization suggests that emerging markets are more susceptible to have procyclical currencies, whereas advanced economies, especially those with refuge currencies, to have countercyclical ones. Second, the degree of procyclicality of a currency affects the policy trade-offs that the countries face. Particularly, the countries with procyclical currencies (and an open capital account) may find it difficult to pursue countercyclical monetary policy, as it would amplify the exchange rate fluctuations. Third, in the last decade countries seem to have become more immune to the procyclicality of their currencies, as their fear of floating seems to have faded. Thus, many emerging markets have adopted more countercyclical monetary policies, often in conjunction with an inflation targeting framework. Our analysis does not provide any a priori reasons for which a country would do better with a procyclical than with a countercyclical currency. On the one hand, if countries with procyclical currency are less able to pursue countercyclical monetary policy, they might be less able to avoid the credit booms that often end up in bust and lead to financial crises. 23 Indeed, we find some preliminary evidence that countries with procyclical currencies tend to be more prone to banking crises, but further analysis is needed to check the robustness of these findings. On the other hand, for export oriented countries with a large manufacturing base, procyclical currency may have the countercyclical property of leading to “competitive devaluations” in bad times and this may be one of the reasons why the adoption of the (countercyclical) euro had created significant problems to the Eurozone periphery countries. Although our analysis does not support any measures aimed at modifying the degree of cyclicality of a currency, we emphasize the fact that the cyclical property of a currency should affect debt management policies. In a country with a procyclical currency, moving from a short to a long net foreign asset position may mean a significant reduction in vulnerability to external shocks. A good example is Brazil, a country with a procyclical currency, which was short in dollars during the previous episodes of financial turmoil. A depreciation of its currency increased its indebtedness and acted as an amplifier to the effect of the financial crises. However, Brazil braced the recent global financial crisis better with a much improved debt position. It was long in dollars at the outset of the crisis and a depreciation of its currency improved its debt position further, creating the much needed space for countercyclical fiscal policies. 23 See, among others, Reinhart and Rogoff (2009), and Schularick and Taylor (2012). 20 References Adler, G., and N. Magud (2013), “Four Decades of Terms of-Trade Booms: A New Metric of Income Windfall and Saving Investment Patterns,” IMF Working Paper 13/103. Aizenman, J., Chinn, M. D., and H. Ito (2013), “The “impossible trinity” hypothesis in an era of global imbalances: Measurement and testing,” Review of International Economics, 21(3): 447–458. Backus, D., and G. Smith (1993), “Consumption and Real Exchange Rates in Dynamics Economies with Non-Traded Goods,” Journal of International Economics 35: 297–316 Benedrix, A., Lane, P. and J.C. Shambaugh (2014) “International Currency Exposures, Valuation Effects and the Global Financial Crisis,” unpublished manuscript. Bluedorn, J., Duttagupta, R., Guajardo, J., and P. (2013), “Capital Flows are Fickle: Anytime, Anywhere,” IMF Working Paper 13/183 Calvo, G. A., and Reinhart, C. M. (2002), “Fear of Floating,”Quarterly Journal of Economics, 117: 379– 408. Cashin, P., Céspedes, L.-F., and R. Sahay (2004), “Commodity currencies and the real exchange rate,” Journal of Development Economics, 75: 239–68 Chari, V. V., Kehoe, P. J., and E. R. McGrattan (2002), “Can Sticky Prices Models Generate Volatility and Persistent Exchange Rates?” Review of Economic Studies, 69: 633–63 Chen, Y., and K. Rogoff (2003), “Commodity currencies,” Journal of International Economics, Vol. 60: 133–160 Chinn, M. (2013), “Macro Approaches to Foreign Exchange Determination,” In J. James, I. W. Marsh and L. Sarno (Eds.) The Handbook of Exchange Rates.” Hoboken NJ: John Wiley & Sons, Inc. Chinn, M. and H. Ito (2006), “What Matters for Financial Development? Capital Controls, Institutions, and Interactions,” Journal of Development Economics, Vol. 81: 163–192. Cordella, T., Federico, P., Vegh, C., and G. Vuletin, (2014), Reserve Requirements in the Brave New Macroprudential World. Washington DC: World Bank. Corsetti, G., Dedola, L., and L. Leduc (2008), “International Risk Sharing and the Transmission of Productivity Shocks,” Review of Economic Studies, 75: 443–73. Dornbusch, R. (1976), “Expectations and Exchange Rate Dynamics,” Journal of Political Economy, 84:1161–76 21 Eichengreen, B., Hausmann, R., and U. Panizza (2007), “Currency mismatches, debt intolerance, and the original sin: Why they are not the same and why it matters,” in Capital controls and capital flows in emerging economies: Policies, practices and consequences (pp. 121–170), Chicago: University of Chicago Press. Engel, C., Mark, N., and K. West (2008), “Exchange Rate Models Are Not as Bad as You Think,” NBER Macroeconomics Annual, 22:381–441 Frenkel, Jacob A. (1976), “A Monetary Approach to the Exchange Rate: Doctrinal Aspects and Empirical Evidence,” Scandinavian Journal of Economics, 78(2): 200–24. Kaminsky, G., Reinhart, C., and C. Végh (2004), “When It Rains, It Pours: Procyclical Capital Flows and Macroeconomic Policies,” NBER Macroeconomics Annual, 19: 11–82. Klein, M. W., and J.C. Shambaugh (2013), “Rounding the Corners of the Policy Trilemma: Sources of Monetary Policy Autonomy,” National Bureau of Economic Research Working Paper 19461. Lane, P., and J.C. Shambaugh (2010), “Financial Exchange Rates and International Currency Exposures,” American Economic Review 100: 518–40. Lederman, D., and W. Maloney (2008), “In Search of the Missing Resource Curse,” World Bank Policy Research Working Paper 4766. Lucas, R. (1982), “Interest rates and Currency prices in a Two-Country World,” Journal of Monetary Economics, 10: 335–59 McGettigan, D., Moriyama, K., Ntsama, J., Painchaud, F., Qu, H. and C. Steinberg (2013). “Monetary Policy in Emerging Markets: Taming the Cycle”, IMF Working Paper, 13/96. Obstfeld, M. and K. Rogoff (1995), “Exchange Rate Dynamics Redux,” Journal of Political Economy, 100: 624–660 Obstfeld, M, and Shambaugh, J. C., and A. M. Taylor (2005), “The trilemma in history: tradeoffs among exchange rates, monetary policies, and capital mobility,” Review of Economics and Statistics, 87(3): 423–438. Reinhart, C. M., and K. Rogoff (2009), This time is different: Eight centuries of financial folly, Princeton NJ: Princeton University Press. Rogoff, K. (2009), “Exchange Rates in the Modern Floating Era: What Do We Really Know?” Review of World Economics 145: 1–12 Schularick, M., and A.M. Taylor (2012), “Credit Booms Gone Bust: Monetary Policy, Leverage Cycles, and Financial Crises, 1870–2008,” American Economic Review, 102: 1029–1061. 22 Vegh, C., and G. Vuletin (2012). Overcoming the fear of free falling: Monetary policy graduation in emerging markets, NBER Working Paper No. 17753 (2012), forthcoming in “The Role of Central Banks in Financial stability: How has it changed?” (Federal Reserve Bank of Chicago). Velculescu, D. (2008), “Norway’s Oil Fund Shows the Way for Wealth Funds,” IMF Survey, July 9. 23 Appendix A Number of observations by country used to compute the “procyclicality” of REER for the entire period Our dataset has 63 countries. Data series go, at best, from 1975q1 to 2013q1 (for quarterly data series), and from 1975 to 2013 (for annual data series). The criterion for the choice of countries is availability of real GDP (in quarterly data). Table A1: Countries included in the data and the data availability Quarterly NEER data Quarterly REER data Quarterly GDP data (from when to when) (from when to when) (from when to when) Argentina 1994q1-2013q1 1994q1-2013q1 1990q1-2012q4 Australia 1979q1-2013q1 1980q1-2013q1 1975q1-2012q4 Austria 1975q1-2013q1 1975q1-2013q1 1975q1-2012q4 Belgium 1975q1-2013q1 1975q1-2013q1 1980q1-2012q4 Belize 1975q1-2013q1 1980q1-2013q1 2000q1-2012q4 Bolivia 1975q1-2013q1 1980q1-2013q1 1990q1-2010q4 Brazil 1975q1-2013q1 1980q1-2013q1 1995q1-2011q4 Bulgaria 1992q1-2013q1 1992q1-2013q1 1999q1-2012q4 Canada 1975q1-2013q1 1975q2-2013q1 1975q1-2013q1 Chile 1975q1-2013q1 1980q1-2013q1 1980q1-2012q4 China 1975q1-2013q1 1980q1-2013q1 2000q1-2011q4 Colombia 1975q1-2013q1 1980q1-2013q1 1994q1-2012q4 Costa Rica 1975q1-2013q1 1980q1-2013q1 2000q1-2011q4 Croatia 1992q1-2013q1 1992q1-2013q1 1993q1-2012q4 Cyprus 1975q1-2013q1 1980q1-2013q1 1995q1-2012q3 Czech Republic 1994q1-2012q4 1990q1-2013q1 1990q1-2013q1 Denmark 1975q1-2013q1 1975q1-2013q1 1977q1-2012q3 Estonia 1994q1-2013q1 1994q1-2013q1 1993q1-2012q3 Finland 1975q1-2013q1 1975q1-2013q1 1975q1-2012q3 France 1975q1-2013q1 1980q1-2013q1 1975q1-2012q3 Georgia 1993q4-2013q1 1993q4-2013q1 1996q1-2011q4 Germany 1975q1-2013q1 1975q1-2013q1 1975q1-2012q3 Greece 1975q1-2013q1 1980q1-2013q1 2001q1-2012q4 Hong Kong SAR, China 1975q1-2013q1 1975q1-2013q1 Hungary 1975q1-2013q1 1980q1-2013q1 1995q1-2012q4 Iceland 1975q1-2013q1 1975q2-2013q1 1997q1-2012q3 India 1994q1-2013q1 1994q1-2013q1 1996q4-2012q3 Indonesia 1994q1-2013q1 1994q1-2013q1 1997q1-2013q1 Iran, Islamic Rep. 1975q1-2013q1 1975q2-2013q1 1988q1-2007q4 Ireland 1975q1-2013q1 1975q2-2013q1 1997q1-2012q3 Israel 1975q1-2013q1 1975q1-2013q1 1975q1-2013q1 24 Italy 1975q1-2013q1 1980q1-2013q1 1980q1-2012q4 Japan 1975q1-2013q1 1980q1-2013q1 1975q1-2012q4 Korea, Rep. 1994q1-2013q1 1994q1-2013q1 1975q1-2012q4 Latvia 1994q1-2013q1 1994q1-2013q1 1990q1-2012q4 Lithuania 1994q1-2013q1 1994q1-2013q1 1993q1-2012q4 Luxembourg 1975q1-2013q1 1975q2-2013q1 1995q1-2013q3 Macedonia, FYR 1992q1-2013q1 1992q1-2013q1 2004q1-2012q4 Malaysia 1975q1-2013q1 1975q1-2013q1 1988q1-2012q4 Malta 1975q1-2013q1 1975q1-2013q1 1996q1-2012q4 Mexico 1975q1-2013q1 1980q1-2013q1 1980q1-2012q4 Morocco 1975q1-2013q1 1980q1-2013q1 1990q1-2012q2 Netherlands 1975q1-2013q1 1975q1-2013q1 1977q1-2012q3 New Zealand 1975q1-2013q1 1975q2-2013q1 1982q2-2012q4 Norway 1975q1-2013q1 1975q1-2013q1 1975q1-2013q1 Peru 1994q1-2013q1 1994q1-2013q1 1979q1-2012q4 Philippines 1975q1-2013q1 1975q2-2013q1 1981q1-2012q4 Poland 1975q1-2013q1 1980q1-2013q1 1995q1-2012q3 Portugal 1975q1-2013q1 1975q1-2013q1 1977q1-2011q4 Romania 1975q1-2013q1 1990q4-2013q1 1998q1-2012q4 Russian Federation 1994q1-2013q1 1994q1-2013q1 1995q1-2012q1 Slovak Republic 1993q1-2012q4 1990q1-2013q1 1990q1-2013q1 Slovenia 1994q1-2013q1 1994q1-2013q1 1992q1-2012q4 South Africa 1975q1-2013q1 1975q1-2013q1 1975q1-2012q4 Spain 1975q1-2013q1 1980q1-2013q1 1975q1-2012q3 Sweden 1975q1-2013q1 1975q1-2013q1 1975q1-2012q4 Switzerland 1975q1-2013q1 1975q1-2013q1 1975q1-2012q4 Thailand 1994q1-2013q1 1994q1-2013q1 1993q1-2013q1 Turkey 1994q1-2013q1 1994q1-2013q1 1987q1-2012q4 Ukraine 1992q1-2013q1 1992q1-2013q1 2001q1-2012q4 United Kingdom 1975q1-2013q1 1975q2-2013q1 1975q1-2012q4 United States 1975q1-2013q1 1980q1-2013q1 1975q1-2013q1 Uruguay 1975q1-2013q1 1980q1-2013q1 2005q1-2011q4 25 Appendix B: Data Sources and Data Details We calculate the currency cyclicality measures CCI(REER/NEER) using quarterly data for GDP and REER/NEER respectively (indexes 2005=100). We also use quarterly capital flow data. All other data series in our final database are annual. • GDP data For the calculation of the currency cyclicality we used real GDP quarterly data (index, 2005=100) from the IMF’s IFS database (annual data is from IFS as well). The cyclical component of the GDP series is derived using a H-P filter with lambda 1600. Only 16 countries had seasonally adjusted data. For the other countries we use the CensusX11 in E-Views to deseasonalize the data series. The availability of data varies across countries, as indicated in Table1. The annual GDP (2005 US$) series are from WDI. • GDP per capita We use GDP and GNI per capita data from WDI, September 2013. We also include GDP in current US$US$ and real GDP (constant 2005 US$USUS$$) from WDI. • World GDP For the calculation of the economy size measure, we used real GDP data from WDI in constant 2005 US$USUS$$. • Economic Size We consider the ratio of real GDP over world GDP per country using data from for domestic GDP and world GDP in constant 2005 US$US$ from WDI. • Exchange rates data The data on Nominal Effective Exchange Rate (NEER) and the Real Effective Exchange Rate (REER) were also from the IFS database; and just like for the GDP series, the cyclical component of the exchange rate series is extracted using the H-P filter. For a few countries, data is complemented with data from the BIS database. Countries for which BIS data are used: Argentina; Hong Kong SAR, China; Estonia; India; Indonesia; the Republic of Korea; Latvia; Lithuania; Peru; Slovenia; Thailand; and Turkey. • Capital Account Openness index We use the index of financial openness developed in Chinn and Ito (2006). The index is based on the binary dummy variables that codify the tabulation of restrictions on cross-border financial 26 transactions reported in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions. The index values range is variable. The last update of the index (2011) gives a score of 2.44 to the “most financially open” country, whereas it gives a score of -1.86 to the “least financial open”. The annual data for this index goes from 1975 to 2011 (http://web.pdx.edu/~ito/Chinn- Ito_website.htm). • De facto exchange rate regime We use the de facto exchange regime classification of Reinhart and Rogoff (http://www.carmenreinhart.com/data/browse-by-topic/topics/11/). The database covers the period 1940-2010 . We use the “coarse” classification (scale 1-6) of de facto exchange rate regime annual data from 1975 to 2010. (-1 is for “strongly fixed” exchange rates; 6 is for “strongly flexible” exchange rates) • Intensity of Net Commodities Exports: We construct a measure of the “intensity” of commodity exports for the different countries in our sample. We define it as net exports of commodities over GDP. Following Lederman and Maloney (2008), we compute net exports of commodities as exports minus imports of natural resources related goods based on Leamer (1995) commodity clusters. Annual data from 1975 to 2011 is taken from the World Integrated Trade Solution (WITS) database. GDP annual data is taken from the IFS database. • Exports, Imports We use the series “Exports of goods and services (% of GDP)”, “Imports of goods and services (% of GDP)”, “Exports of goods and services (current US$)” and “Imports of goods and services (current US$)” from WDI, September 2013. • Capital flows We Bluedon et al. (2013) database, which e includes data on quarterly gross inflows, net capital inflows and GDP in current US$. The data range from 1970 Q1 to 2012Q2. The capital flow cyclicality is calculated the same way as the currency cyclicality. • Monetary and Fiscal Database We use the databases on Monetary and Fiscal procyclicalities by Vegh and Vuletin that accompany their paper: Vegh, C., and G. Vuletin, .Overcoming the fear of free falling: Monetary policy graduation in emerging markets, NBER Working Paper No. 17753 (2012), forthcoming in “The Role of Central Banks in Financial stability: How has it changed?” (Federal Reserve Bank of Chicago). 27 Table B1: Summary statistics of the procyclicality measures Mean Median Min Max CCINEER 0.099 0.041 -0.422 0.727 CCIREER 0.107 0.041 -0.460 0.912 Table B2: Summary statistics of the Dependent Variables Quarterly/ Variable Annual Mean Std. Dev. Min Max Observations Countries NEER (index 5.14E+1 9.95E+1 2.81E+1 2005=100, seas. adj.) quarterly 0 1 3.30 3 8214 63 REER (index 2005=100, seas. adj.) quarterly 107.51 62.36 20.84 1149.61 7769 62 Real GDP (index 2005=100, seas. adj.) quarterly 82.50 24.79 13.25 196.28 6467 63 World GDP (index 2005=100, seas. adj.) quarterly 72.23 26.04 35.51 123.49 152 NEER (index 5.32E+1 1.01E+1 2.78E+1 2005=100) annual 0 2 3.98 3 1982 63 REER (index 2005=100) annual 107.73 62.57 37.51 1123.83 1843 63 Real GDP (index 2005=100) annual 73.89 26.67 8.26 187.91 2101 63 GDP (constant 2005 5.44E+1 1.38E+1 2.09E+0 1.35E+1 US$US$) annual 1 2 8 3 2192 63 World GDP (index 2005=100) annual 72.61 25.68 36.87 123.23 38 World GDP (constant 2005 3.41E+1 1.06E+1 1.85E+1 5.37E+1 US$US$) annual 3 3 3 3 38 Capital Account Openness annual 1.01 1.46 -1.86 2.44 1342 61 Commodity Exporters annual 0.00 0.01 -0.02 0.03 1386 62 Monetary Procyclicality annual -0.23 0.27 -0.68 0.46 1238 41 Fiscal Procyclicality annual -0.07 0.29 -0.71 0.52 1441 42 Total Net Private Capital Inflow (US$, millions) quarter 1,513.36 18,588.68 -250,020 447,450 6,288 52 28 Appendix C Table C1: Currency Cyclicality across Countries Quarterly data Annual Data Two tailed One tailed two tailed test (0.10) test (0.15) (0.10) two tailed test (0.10) CCREER CCNEER CCREER CCREER CCREER Argentina* Argentina* Argentina* Argentina* Argentina* Australia Australia* Australia Australia* Australia Austria Austria Austria+ Austria+ Austria Belgium Belgium Belgium Belgium+ Belgium Belize Belize Belize Belize Belize+ Bolivia Bolivia Bolivia Bolivia Bolivia Brazil* Brazil* Brazil Brazil Brazil Bulgaria Bulgaria+ Bulgaria Bulgaria Bulgaria Canada Canada Canada Canada Canada Chile* Chile Chile Chile Chile China China China China Mainland Mainland Mainland Mainland China Mainland Colombia Colombia Colombia Colombia Colombia Costa Rica+ Costa Rica Costa Rica+ Costa Rica+ Costa Rica Croatia Croatia Croatia Croatia Croatia Cyprus+ Cyprus Cyprus+ Cyprus+ Cyprus Czech Repub Czech Repub Czech Repub Czech Repub Czech Repub Denmark Denmark Denmark Denmark Denmark Estonia+ Estonia+ Estonia+ Estonia+ Estonia Finland Finland Finland Finland Finland France+ France+ France+ France France+ Georgia Georgia Georgia Georgia Georgia Germany+ Germany+ Germany+ Germany+ Germany Greece Greece Greece Greece Greece Hungary Hungary Hungary Hungary Hungary+ Iceland Iceland Iceland Iceland Iceland India India India India India Indonesia Indonesia Indonesia Indonesia Indonesia Iran Iran Iran Iran Iran+ Ireland Ireland+ Ireland Ireland Ireland Israel Israel Israel Israel Israel Italy Italy Italy Italy Italy Japan+ Japan+ Japan+ Japan+ Japan+ Korea Rep Korea Rep Korea Rep Korea Rep Korea Rep Latvia Latvia+ Latvia Latvia+ Latvia Lithuania+ Lithuania+ Lithuania+ Lithuania+ Lithuania Luxembourg Luxembourg+ Luxembourg Luxembourg Luxembourg Macedonia Macedonia Macedonia Macedonia Macedonia Malaysia Malaysia Malaysia Malaysia Malaysia Malta Malta Malta Malta Malta Mexico Mexico Mexico Mexico Mexico Morocco Morocco Morocco Morocco Morocco Netherlands+ Netherlands+ Netherlands+ Netherlands+ Netherlands+ New Zealand New Zealand New Zealand New Zealand New Zealand 29 Norway+ Norway Norway+ Norway+ Norway+ Peru Peru Peru Peru Peru Philippines Philippines Philippines Philippines Philippines Poland Poland Poland Poland Poland Portugal Portugal Portugal Portugal Portugal Romania Romania Romania Romania Romania Russia Russia Russia Russia Russia Slovakia Slovakia Slovakia Slovakia Slovakia Slovenia Slovenia Slovenia Slovenia Slovenia South Africa South Africa South Africa South Africa South Africa Spain Spain Spain Spain Spain Sweden Sweden Sweden Sweden Sweden Switzerland+ Switzerland+ Switzerland+ Switzerland+ Switzerland+ Thailand Thailand Thailand Thailand Thailand Turkey Turkey Turkey Turkey Turkey Ukraine Ukraine Ukraine Ukraine Ukraine United United United United Kingdom Kingdom Kingdom Kingdom United Kingdom United United States States+ United States United States United States Uruguay Uruguay Uruguay Uruguay Uruguay Note: Countries with procyclical currencies are denoted by * (and in red); countries with countrecyclcial currencies are denoted by + (and in blue). 30