Financial Ampli cation of Foreign Exchange Risk Premia 1

Transcription

1 Financial Ampli cation of Foreign Exchange Risk Premia 1 Tobias Adrian, Erkko Etula, Jan Groen Federal Reserve Bank of New York Brussels, July 23-24, 2010 Conference on Advances in International Macroeconomics Lessons from the Crisis ECFIN/ULB/UBC 1 The views expressed in this paper are those of the authors and do not necessarily re ect the position of the Federal Reserve Bank of New York or the Federal Reserve System.

2 Financial Market Monitoring I Are current asset valuations "high" or "low"? I How do the valuations link to macro fundamentals? I What are the implications for nancial stability?

3 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001):

4 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001): I Interaction of constraints on the domestic and foreign nancial sectors lead to ampli cation of macroeconomic shocks

5 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001): I Interaction of constraints on the domestic and foreign nancial sectors lead to ampli cation of macroeconomic shocks I Brunnermeier and Pedersen (2009):

6 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001): I Interaction of constraints on the domestic and foreign nancial sectors lead to ampli cation of macroeconomic shocks I Brunnermeier and Pedersen (2009): I Funding liquidity and market liquidity interact, leading to spillovers across nancial institutions and excessively volatile asset prices

7 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001): I Interaction of constraints on the domestic and foreign nancial sectors lead to ampli cation of macroeconomic shocks I Brunnermeier and Pedersen (2009): I Funding liquidity and market liquidity interact, leading to spillovers across nancial institutions and excessively volatile asset prices I Korinek (2010):

8 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001): I Interaction of constraints on the domestic and foreign nancial sectors lead to ampli cation of macroeconomic shocks I Brunnermeier and Pedersen (2009): I Funding liquidity and market liquidity interact, leading to spillovers across nancial institutions and excessively volatile asset prices I Korinek (2010): I Ties systemic risk in international capital ows explicitly to the welfare losses due to balance sheet constraints of nancial institutions

9 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001): I Interaction of constraints on the domestic and foreign nancial sectors lead to ampli cation of macroeconomic shocks I Brunnermeier and Pedersen (2009): I Funding liquidity and market liquidity interact, leading to spillovers across nancial institutions and excessively volatile asset prices I Korinek (2010): I Ties systemic risk in international capital ows explicitly to the welfare losses due to balance sheet constraints of nancial institutions I Veld, Roeger, Raciborski, Ratto (2010), Devreux and Yetman (2010), Quadrini and Perri (2010), Enders, Kollmann, Mueller (2010), Tille and Von Wincoop (2010):

10 Systemic Risk in the FX Markets I Caballero and Krishnamurthy (2001): I Interaction of constraints on the domestic and foreign nancial sectors lead to ampli cation of macroeconomic shocks I Brunnermeier and Pedersen (2009): I Funding liquidity and market liquidity interact, leading to spillovers across nancial institutions and excessively volatile asset prices I Korinek (2010): I Ties systemic risk in international capital ows explicitly to the welfare losses due to balance sheet constraints of nancial institutions I Veld, Roeger, Raciborski, Ratto (2010), Devreux and Yetman (2010), Quadrini and Perri (2010), Enders, Kollmann, Mueller (2010), Tille and Von Wincoop (2010): I International macro and leverage dynamics

11 What we do I Extract the common components of expected U.S. dollar funded carry trade returns via partial least squares regressions I Estimate foreign exchange risk premia: 1. Macro fundamentals (real activity and in ation) 2. Balance sheet quantities ("funding liquidity conditions")

12 Risk Premium Kernel Density Estimates Macro Variables Only Macro and Balance Sheet Variables

13 Macro and Balance Sheet Ampli cation Macro

14 Episodes of "Balance Sheet Risk Premium" Compression 1. From 1988 (Louvre Accord) to early 1991 (global downturn) 2. Run-up to the LTCM crisis, between early 1995 and Balance sheet risk premium exhibited a prolonged decline between July 2002 and June 2008 I Each of these episodes of prolonged risk premium compression is followed by a sharp reversal of the balance sheet risk premium in the crisis

15 Related Literature 1. Violation of UIP (Fama 1984, Engel 1996 among others) 2. Common risk factors in carry returns (Mahieu and Schotman 1994, Lustig, Roussanov and Verdelhan 2010) 3. Asset pricing implications of funding liquidity risk (Adrian and Shin 2007, Adrian and Etula 2010, Adrian, Moench and Shin 2010)

16 Outline 1. State Variable Extraction via Partial Least Squares 2. Risk Premium Estimation 3. Implications for Systemic Risk Monitoring

17 Data I Monthly data: 1. Exchange rates, 2. Macroeconomic fundamentals, 3. Aggregate balance sheet components of U.S. nancial institutions. I January 1988 to December 2010 I Start dictated by the availability of balance sheet data.

18 Foreign Exchange Return I Riskless foreign bond holding period rate of return r i f,t, I U.S. dollar funding is riskless at rate r US f,t, I Spot exchange rate, ε i t: number of U.S. dollars that can be bought with one unit of foreign currency i, I The excess return to this strategy is given by: ert+1 i 1 + rf i ε i t+1,t ε i t 1 + r US f,t I Data from Datastream I Australia, Austria, Belgium, Canada, Hong Kong, Czech Republic, Denmark, Euro area, Finland, France, Germany, Greece, Hungary, India, Indonesia, Ireland, Italy, Japan, Kuwait, Malaysia, Mexico, Netherlands, New Zealand, Norway, Philippines, Poland, Portugal, Saudi Arabia, Singapore, South Africa, South Korea, Spain, Sweden, Switzerland, Taiwan, Thailand, United Kingdom

19 Currency Portfolios I We form six currency portfolios on the basis of the forward premium at the end of each month I We sort the currencies according to the forward premium vis-à-vis the U.S. dollar each month I Compute the excess return for each of six forward premium portfolios. This provides a way to: I deal with the unbalanced panel of the currency data I increase the stability of the risk factor loadings

20 Real Activity Panel I Panel consists of 44 real activity series, including: I I I I I Industrial production data for the United Kingdom, Denmark, France, Germany (both total and excluding construction), Spain, Austria, Belgium, Italy, Luxembourg, Norway, Ireland, Portugal, Taiwan, India, Korea, Malaysia, United States, Japan, and Capacity utilization rate data for the manufacturing sectors in Japan and the United States, Consumer and business con dence indicators for the euro area, France, Italy, Netherlands, the European Union, and the United States, Business con dence indicators for the United Kingdom, Austria, Belgium, Denmark, Luxembourg, Finland, Greece, and Portugal, Consumer con dence indicator for Spain. I We use annual growth rates for industrial production series.

21 In ation Panel I Panel consists of annual consumer price index (CPI) in ation data for 26 economies, i.e., the United States, the United Kingdom, Belgium, Denmark, France, Germany, Italy, Norway, Sweden, Switzerland, Canada, Japan, Finland, Greece, Ireland, Portugal, Spain, Taiwan, Hong Kong, Korea, Malaysia, Pakistan, Philippines, Singapore, Thailand, China. I We take the 12-month di erence in annual (year-on-year) CPI in ation rates.

22 Aggregate Balance Sheet Components 1. U.S. dollar nancial commercial paper (CP) outstanding cleared at the Depository Trust and Clearing Corporation. I Includes USD CP issued by U.S. and foreign nancial institutions with U.S. a liates. 2. Bond issues of U.S nancial corporations and non- nancial corporations. I We take the logarithm of the ratio of nancial bonds issued relative to non- nancial bonds issued each month. 3. Free credit balances at U.S. broker-dealer margin accounts. 4. Debit balances at U.S. broker-dealer margin accounts.

23 Balance Sheet Variables (Standardized) 1990m1 1995m1 2000m1 2005m1 2010m1 Financial Commercial Paper Debit Balances at Broker Dealer Margin Accounts Free Credit Balances at Broker Dealer Margin Accounts Financials Bond Issues Relative to Non Financials Bond Issues

24 Partial Least Squares I I PLS is an e cient factor estimation approach (Groen and Kapetanios, 2008). I T N matrix of N indicator variables Z = (z 0 1 z 0 T )0 (consisting of real activity, in ation and balance sheet data). I Standardize Z to have zero mean and unit variance space, resulting in the T N matrix Z = ( z 0 1 z 0 T )0. I PLS uses dominant eigenvector v of the squared covariance between the vector of demeaned dollar-based currency returns and the panel of combined predictor variables: Z 0 er er 0 Z where er = ( er 10 er 60 ) 0 with er i = ( er i 1 er i T )0 where is the demeaned excess return on currency i. er i t

25 Partial Least Squares II I The common factor from Z relevant for the dollar-based excess returns is then: X t = ( v z t ) 0 where v is a transformation of the N 1 dominant eigenvector v of Z 0 er er 0 Z such that jjvjj = 1. I Then we decompose X t into a global real activity subfactor Xt real, a global in ation subfactor Xt in, and an aggregate U.S. balance sheet subfactor Xt BS I we impose a hierachical factor structure where X t corresponds to a linear combination of Xt real, Xt in, Xt BS. I This is implemented via an iterative process by extracting linear combination of X s that correlate most highly with the common component of the excess currency returns.

26 Real Activity Inflation Cum. Carry Ret Balance Sheet Estimated State Variables Cumulative Average Carry Return Balance Sheet State Variable 1990m1 1995m1 2000m1 2005m1 2010m1 1990m1 1995m1 2000m1 2005m1 2010m1 Real Activity State Variable Inflation State Variable 1990m1 1995m1 2000m1 2005m1 2010m1 1990m1 1995m1 2000m1 2005m1 2010m1

27 Asset Pricing Approach I I With a pricing kernel M t+1 /M t, the expected payo is: " M t+1 E t 1 + r i ε i!# t+1 f,t M t ε i 1 + rf US,t = 0 t I Using the de nition of covariance, we nd the uncovered interest rate parity: ε i t+1 ε i t = 1 + r US f,t 1 + r i f,t + µ t + ξ i t+1 where ξ i t+1 denote exchange rate innovations and µ t foreign exchange risk premium: " # M t+1 /M t µ t = Cov t E t [M t+1 /M t ], εi t+1 ε i t is the

28 Asset Pricing Approach II I It follows that exchange rate appreciation is due to three components: ε i t+1 ε i t {z} Exchange Rate Appreciation = 1 + r US f,t 1 + r i f,t {z } Interest Rate Carry + µ t {z} FX Risk Premium + ξ i t+1. {z} FX Risk

29 Empirical Implementation I I We assume that the pricing kernel M t+1 /M t is exponentially a ne in the state variables X t : M t+1 = exp r f 1 t M t 2 λ0 t λ t λtv 0 t+1 Σ t λ t = λ 0 + λ 1 X t where X t+1 = µ + φx t + v t+1. I We further assume that the innovations to state variables are normally distributed with v t+1 N (0, Σ t ).

30 Empirical Implementation II I With this notation, we use Stein s lemma to express the FX risk premium as: " # M t+1 /M t µ t = Cov t E t [M t+1 /M t ], εi t+1 = Cov t " v t+1, εi t+1 ε i t # ε i t Σ 1 t (λ 0 + λ 1 X t ). I It follows that the pricing equation reduces to: where β i0 t ε i t+1 ε i t = 1 + r US f,t 1 + r i f,t h = Cov t v t+1, 1/εi t+1 + β i0 t (λ 0 + λ 1 X t ) + ξ i t+1, 1/ε i t i Σ 1 t.

31 Empirical Implementation III I The exchange rate risk ξ i t+1 can further be decomposed into a systematic component β i0 t v t+1, and an idiosyncratic component et+1 i, such that: ε i t+1 ε i t {z} FX Appreciation 1 + r US f,t 1 + r i f,t {z } Carry = β i0 t (λ 0 + λ 1 X t ) + {z } β i0 t v t+1 {z } FX Risk Systematic Premia FX Risk + e i t+1 {z}. Idiosyncratic FX Risk

32 Empirical Implementation IV I We next investigate the extent to which the state variables determine the FX risk premium. I We de ne the systematic FX risk as the unforecastable part of the return to an equal-weighted carry portfolio: v t+1 = r EW t+1 E t h r EW t+1 j1, rt EW, X t i, I We estimate cross-sectional model by way of three-step OLS regressions applied to the cross-section of 6 carry portfolios (see Adrian and Moench, 2008). I For simplicity, we estimate constant betas for each portfolio i.

33 Risk Premium of the Equal Weighted Carry Portfolio 1990m1 1995m1 2000m1 2005m1 2010m1 Foreign Exchange Returns Risk Premium

34 Risk Premium of Macro and Balance Sheet Variables 1990m1 1995m1 2000m1 2005m1 2010m1 Macro and Balance Sheet Risk Premium Macro Risk Premium

35 Macro and Balance Sheet Ampli cation Macro

36 Risk Premium Kernel Density Estimates Macro Variables Only Macro and Balance Sheet Variables Figure:

37 Implications for Financial Stability Monitoring Systemic risk regulators: I Monitor the evolution of risk in the nancial system, I Develop models to detect the buildup of potential vulnerabilities, and I Formulate appropriate policies. This paper presents a methodology to measure the risk premium associated with the dynamics of intermediary balance sheets, which in turn is an indicator for the buildup of systemic risk. I We document that the nancial intermediary balance sheet variables amplify the volatility of the FX risk premium relative to the risk premium due to macro fundamentals.

38 FX Volatility and the Balance Sheet Risk Premium I FX volatility is an indicator of risk I Episodes of systemic nancial instability are usually accompanied with high FX volatility I However, high FX volatility does not necessarily correspond to systemic risk.

39 FX Volatility and the Balance Sheet Risk Premium 1990m1 1995m1 2000m1 2005m1 2010m1 Standardized Log FX Volatility Standardized Balance Sheet Risk Premium

40 Conclusion I We document that the presence of balance sheet variables substantially ampli es the volatility of estimated foreign exchange risk premiums. I This nding is in line with theories of systemic risk that argue that balance sheet constraints lead to excess volatility of exchange rates and asset prices more generally. I The methodology that we develop can be used for quantitative surveillance systems to monitor nancial stability.