Currency Premia and Global Imbalances Conference on Macro-Financial Linkages & Current Account Imbalances,Vienna Pasquale Della Corte Steven J. Riddiough Lucio Sarno Imperial College London University of Warwick Cass Business School July 2015 Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 1 / 22
Introduction A carry trade investor earns an excess return by selling low-yielding currencies and investing in high-yielding currencies. but the risk is large as every once in a while, investment currencies suddenly depreciate thus causing large losses A risk-based view would suggest that investment currencies o er a premium for higher risk exposure (Fama, 1984). Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 2 / 22
Motivations A recent literature provides evidence that carry trade returns can be understood as compensation for global risk Lustig, Roussanov and Verdelhan (2011) propose a slope factor as a proxy for global risk Menkho, Sarno, Schmeling and Schrimpf (2012) use innovations to the average foreign exchange volatility. What economic forces drive global risk? the existing factors can be considered as proximate risk factors as they are based on nancial variables, in theory, determined by economic fundamentals in this paper we tackle this issue and shed light on the economic determinants of currency risk premia. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 3 / 22
Overview What we do... guided by a model of capital ows in imperfect nancial markets (Gabaix and Maggiori, 2015) we test whether external imbalances can explain carry trade returns in a standard asset pricing framework.... our main results we identify a global imbalance risk factor that empirically explains the average carry trade returns, and denote it as IMB we thus support a risk-based interpretation of currency excess returns using economic fundamentals Gourinchas and Rey (2007), Gourinchas (2008), Habib and Stracca (2012), and Catão and Milesi-Ferretti (2014), among many others. F A carry trade investor demands compensation for his exposure to global imbalance risk. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 4 / 22
Theoretical Motivations Gabaix and Maggiori (2015) propose a two-country model where each country borrows or lends in its own currency, intermediaries demand compensation (expected currency appreciation) for holding currency risk arising from imbalanced capital ows exchange rates are jointly determined by global imbalances and intermediaries risk-bearing capacity. Carry trade returns E (imp 1 ) E (RX ) = Γ i i E (imp 1 ) imp 0 (i + Γ) imp 0 + i i E (imp 1 ) imp 0 determines net foreign asset positions, i and i are the domestic and foreign riskless interest rates, Γ captures the risk-bearing capacity of nanciers Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 5 / 22
Testable Predictions Hypothesis 1: the carry trade return is bigger when the interest rate di erential is larger, the investment (funding) country is a net foreign debtor (creditor), and the investment (funding) country has a higher propensity to issue liabilities in foreign (domestic) currency (see Gourinchas, 2008). Hypothesis 2: when there is a nancial disruption ( Γ increases), net debtor countries experience a currency depreciation, while the opposite is true for net-creditor countries. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 6 / 22
List of Countries The All Countries sample (55): Argentina, Australia, Austria, Belgium, Brazil, Canada, Chile, China, Colombia, Croatia, Czech Republic, Denmark, Egypt, Estonia, Euro Area, Finland, France, Germany, Greece, Hong Kong, Hungary, Iceland, India, Indonesia, Ireland, Israel, Italy, Japan, Kazakhstan, Latvia, Lithuania, Malaysia, Mexico, Morocco, Netherlands, New Zealand, Norway, Philippines, Poland, Portugal, Russia, Singapore, Slovakia, Slovenia, South Africa, South Korea, Spain, Sweden, Switzerland, Thailand, Tunisia, Turkey, Ukraine, United Kingdom, and Venezuela. The Developed Countries sample (15): highlighted in blue. The sample ranges from October 1983 to June 2014 the number of available currencies changes over time. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 7 / 22
Data Foreign exchange rates: 1983 2014 end-of-month spot and 1-month forward exchange rates, source: Barclays and Reuters data via datastream, de nition: units of USD per unit of foreign currency. External imbalance data: 1982 2013 annual data on the net foreign asset position to GDP ratio (nfa), source: Lane and Milesi-Ferretti (2007) and the IFS database, we construct monthly data by forward lling. Currency denomination of foreign liabilities: 1990 2012 yearly data on the fraction of foreign liabilities in local currency (ldc), souce: Benetrix, Lane and Shambaugh (2014), we construct monthly data by backward/forward lling. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 8 / 22
Carry Trade Portfolios At time t, group currencies into 5 portfolios using their forward discounts fd t P 1 : low-yielding currencies (funding currencies) P 5 : high-yielding currencies (investment currencies) At time t + 1, each portfolio s excess return is computed as the average individual excess return. Lustig, Roussanov, and Verdelhan (2011) the rst principal component implies an equally weighted strategy across all portfolios, and de nes the DOL factor the second principal component is equivalent to a long-short strategy between P 5 and P 1, and is denoted as CAR factor. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 9 / 22
Global Imbalance Portfolios At time t, sort currencies into 2 baskets using nfa t re-order each basket using ldc t P 1 : creditor countries with liabilities mainly in domestic currency P 5 : debtor countries with liabilities primarily in foreign currency. At time t + 1, each portfolio s excess return is computed as the average individual excess return. Global imbalance factor we construct the imbalance factor as P 5 minus P 1, and refer to it as the IMB factor. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 10 / 22
Global Imbalance Portfolios Sorting Scheme Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 11 / 22
Descriptive Statistics Global Imbalance Portfolios P 1 P 2 P 3 P 4 P 5 IMB All Countries mean 0.92 3.51 1.40 3.57 5.32 4.40 t-stat [0.60] [2.18] [1.10] [2.39] [2.73] [3.51] sdev 7.80 8.71 6.52 7.92 10.05 6.43 SR 0.12 0.40 0.22 0.45 0.53 0.68 mdd 0.46 0.29 0.33 0.26 0.30 0.20 fd 0.54 1.20 2.02 3.50 6.80 nfa 0.43 0.14 0.10 0.46 0.56 ldc 0.63 0.47 0.44 0.47 0.28 Developed Countries mean 0.67 2.45 3.06 3.46 4.65 3.98 t-stat [0.37] [1.31] [1.77] [2.00] [2.38] [3.26] sdev 9.90 10.25 9.33 9.06 10.29 6.76 SR 0.07 0.24 0.33 0.38 0.45 0.59 mdd 0.54 0.36 0.34 0.32 0.31 0.26 fd 1.32 0.76 1.81 2.15 2.23 nfa 0.41 0.31 0.04 0.37 0.37 ldc 0.61 0.46 0.48 0.49 0.34 Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 12 / 22
Excess Returns and External Imbalances distribution of the global imbalance factor conditional on carry trade returns Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 13 / 22
Asset Pricing Tests: Methods The Euler condition E [M t RX j t ] = 0 RXt j : excess return on portfolio j M t = 1 b 0 f t : linear stochastic discount factor b: vector of factor loadings f t : demeaned risk factors The β-pricing model λ: market price of risk E [RX j ] = Cov(RX j t, f t ) Var (f t ) {z } β j β j : regression coe cient of RX j t+1 on f t+1. bvar (f t ) {z } λ Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 14 / 22
Asset Pricing Tests: Pricing Carry Trade Portfolios Cross-Section λ DOL λ IMB R 2 HJ λ DOL λ IMB R 2 HJ All Countries Developed Countries GMM 1 0.02 0.07 0.87 0.14 0.02 0.05 0.91 0.06 (0.01) (0.02) [0.16] (0.02) (0.02) [0.80] GMM 2 0.02 0.07 0.74 0.02 0.05 0.88 (0.01) (0.02) (0.02) (0.02) FMB 0.02 0.07 0.87 0.02 0.05 0.91 (0.01) (0.02) (0.02) (0.02) Time-Series α β DOL β IMB R 2 α β DOL β IMB R 2 P 1 0.01 1.00 0.33 0.80 0.01 0.97 0.46 0.74 (0.01) (0.05) (0.04) (0.01) (0.04) (0.06) P 2 0.02 0.99 0.17 0.83 0.01 1.01 0.16 0.82 (0.01) (0.04) (0.03) (0.01) (0.04) (0.04) P 3 0.01 1.05 0.10 0.85 0.01 0.97 0.01 0.86 (0.01) (0.03) (0.02) (0.01) (0.03) (0.03) P 4 0.01 1.04 0.12 0.82 0.01 0.97 0.14 0.83 (0.01) (0.04) (0.05) (0.01) (0.03) (0.04) P 5 0.01 0.90 0.46 0.74 0.01 1.02 0.52 0.77 (0.01) (0.05) (0.08) (0.01) (0.04) (0.06) HJ: test for the null hypothesis of zero pricing errors with p-values in parentheses Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 15 / 22
Summary Statistics: Beta-sorted Portfolios P 1 P 2 P 3 P 4 P 5 H /L All Countries mean 0.54 2.18 3.85 3.10 4.67 5.21 t-stat [ 0.38] [1.49] [2.39] [1.59] [2.38] [2.83] sdev 6.62 7.62 8.18 9.10 9.61 9.11 SR 0.08 0.29 0.47 0.34 0.49 0.57 mdd 0.49 0.35 0.18 0.30 0.26 0.20 nfa 0.45 0.03 0.02 0.11 0.41 ldc 0.53 0.50 0.48 0.46 0.41 fd 0.36 0.55 2.13 2.60 4.30 pre-β 0.22 0.14 0.51 0.78 1.35 post-β 0.30 0.31 0.09 0.07 0.31 Developed Countries mean 1.02 3.61 2.47 2.33 4.92 5.93 t-stat [ 0.51] [1.80] [1.31] [1.40] [2.33] [2.76] sdev 9.74 10.29 9.25 8.51 10.59 10.79 SR 0.10 0.35 0.27 0.27 0.46 0.55 mdd 0.65 0.36 0.30 0.27 0.33 0.42 nfa 0.47 0.28 0.04 0.18 0.49 ldc 0.57 0.49 0.47 0.46 0.46 fd 1.53 0.02 0.89 1.45 3.04 pre-β 0.94 0.50 0.28 0.05 0.67 post-β 0.57 0.19 0.03 0.12 0.59 Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 16 / 22
Asset Pricing Tests: Double-Sorted Portfolios NFA LDC R 2 χ 2 HJ All Countries b GMM 1 1.10 b 1.56 c 0.80 6.28 0.17 λ GMM 1 0.02 b 0.04 c b GMM 2 0.85 b 1.41 c 0.78 5.65 λ GMM 2 0.02 b 0.05 c b FMB 1.09 c 1.56 c 0.80 6.28 λ FMB 0.02 b 0.05 c D-test [0.03] [<.01] Developed Countries b GMM 1 0.70 a 0.63 0.87 2.51 0.09 λ GMM 1 0.02 b 0.02 b b GMM 2 0.59 a 0.55 0.92 2.35 λ GMM 2 0.03 b 0.01 b FMB 0.69 b 0.63 a 0.87 2.49 λ FMB 0.02 b 0.02 b D-test [0.04] [0.12] Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 17 / 22
Minimum Variance Portfolios: The Value added of the Global Imbalance Factor Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 18 / 22
Fixed-E ects Panel Regressions: Determinants of Spot Returns nominal exchange rate returns (t) (1) (2) (3) (4) (5) (6) nfa t 12 0.043 0.040 0.015 0.158 b 0.159 b 0.113 VIX t 0.143 c 0.143 c 0.126 c VIX t nfa t 12 0.069 c 0.069 c 0.058 c ldc t 12 0.092 0.327 0.027 0.105 fd t 1 0.004 b 0.001 VIX t fd t 1 0.001 VIX dummy t 1.119 c 1.119 c 0.903 c VIX dummy t nfa t 12 0.731 c 0.731 c 0.563 c VIX dummy t fd t 1 0.012 b Adjusted R 2 0.08 0.08 0.08 0.02 0.02 0.02 VIX dummy t = 1 if VIX > σ full sample (and 0 otherwise). Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 19 / 22
Asset Pricing Tests: Other Currency Strategies Portfolios Sample b DOL b IMB λ DOL λ IMB R 2 HJ 5 Carry Trade 10/83 06/14 0.07 1.53 0.02 0.07 0.87 0.14 (0.29) (0.54) (0.01) (0.02) [0.16] 5 Global Imbalance 10/83 06/14 0.30 0.63 0.03 0.04 0.75 0.17 (0.26) (0.29) (0.01) (0.01) [0.92] 5 Momentum 10/83 06/14 0.31 0.22 0.02 0.02 <.01 0.16 (0.30) (0.75) (0.01) (0.03) [0.04] 5 Value 10/83 06/14 0.01 1.33 0.02 0.07 0.66 0.14 (0.30) (0.53) (0.01) (0.02) [0.18] 5 Risk Reversal 01/96 06/14 0.12 1.55 0.02 0.09 0.96 0.04 (0.44) (0.93) (0.02) (0.05) [0.97] 20 Currency 10/83 06/14 0.12 0.97 0.02 0.05 0.53 0.81 (0.27) (0.36) (0.01) (0.01) [0.17] 25 Currency 01/96 06/14 0.08 1.38 0.02 0.08 0.65 0.93 (0.41) (0.38) (0.01) (0.01) [0.86] Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 20 / 22
Asset Pricing Tests: Bond, Currency and Equity Strategies Panel A: Size and Book-to-Market Global Portfolios (SBM) Sample λ MKT λ SMB λ HML λ DOL λ IMB R 2 HJ 20 Cur + 25 SBM 90-14 0.06 0.02 0.07 0.02 0.07 0.78 0.96 (0.02) (0.01) (0.01) (0.01) (0.01) [0.38] 25 Cur + 25 SBM 96-14 0.07 0.01 0.08 0.02 0.09 0.84 1.02 (0.02) (0.01) (0.01) (0.01) (0.01) [0.20] Panel B: Size and Momentum Global Portfolios (SMM) Sample λ MKT λ SMB λ WML λ DOL λ IMB R 2 HJ 20 Cur + 25 SMM 90-14 0.09 0.04 0.10 0.02 0.07 0.83 0.98 (0.02) (0.01) (0.02) (0.01) (0.01) [0.34] 25 Cur + 25 SMM 96-14 0.10 0.04 0.11 0.02 0.09 0.86 1.02 (0.03) (0.01) (0.02) (0.01) (0.01) [0.22] Panel C: International Bond Portfolios (IB) Sample λ IB λ DOL λ IMB R 2 HJ 20 Cur + 5 IB 83-14 0.06 0.02 0.05 0.63 0.92 (0.01) (0.01) (0.01) [0.48] 25 Cur + 5 IB 96-14 0.08 0.02 0.07 0.64 1.02 (0.01) (0.01) (0.01) [0.26] Panel D: Commodity Portfolios (COM) Sample λ COM λ DOL λ IMB R 2 HJ 20 Cur + 7 COM 83-98 0.09 0.02 0.06 0.37 0.89 (0.04) (0.01) (0.01) [0.59] 25 Cur+ 7 COM 96-98 0.15 0.02 0.09 0.61 0.96 (0.03) (0.01) (0.01) [0.41] Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 21 / 22
Conclusions The massive devaluation of high-yielding currencies in the aftermath of Lehman Brothers collapse has revived interest in the risk-return pro le of carry trade returns. If high interest rate currencies deliver low returns in bad times, then currency excess returns compensate investors for higher risk-exposure and UIP deviations re ect time-varying risk premia (Fama, 1984; Engel, 1984). Guided by Gabaix and Maggiori (2015), we construct an external imbalance risk factor that explains the majority of average excess returns in a standard asset pricing model. Debtor countries o er a currency risk premium to compensate investors willing to fund persistent current account de cits. Della Corte, Riddiough & Sarno (2015) Currency Premia & Global Imbalances July 2015 22 / 22