International Liquidity and Exchange Rate Dynamics

Transcription

1 International Liquidity and Exchange Rate Dynamics Xavier Gabaix and Matteo Maggiori New York University, Stern School of Business, NBER, CEPR May Macro Financial Modeling Meeting

2 Imperfect Finance and the Determination of Exchange Rates One very important and quite robust insight is that the nominal exchange rate must be viewed as an asset price Obstfeld and Rogoff (1996) Exchange rates are disconnected from traditional macroeconomic fundamentals They are instead connected to financial forces: e.g. capital flows and financial conditions Demand and Supply of assets in different currencies is central to exchange rate determination Financial determination in imperfect capital markets is key for welfare analysis: floating exchange rates do not move to absorb real shocks as in Mundellian analysis Important issues: framework is desirable, but has proven elusive

3 Imperfect Finance and the Determination of Exchange Rates We provide a basic framework of capital flows and exchange rates: Capital flows alter balance sheet of financiers who intermediate flows Financiers balance sheets and risk bearing capacity determine the required compensation for intermediating capital flows Such compensation determines both the level and dynamics of exchange rates Practical Example: US investors demand Brazilian Real bonds Financiers provide these bonds in the short-medium run, Short Real and Long Dollar To compensate financiers, the Real appreciates on impact and is expected to depreciate relative to the Dollar Our framework is a basic theory where a price, the exchange rate, has to move to balance the demand/supply of assets in financial markets

4 Building up the Framework Basic exchange rate determination in an imperfect financial world US HOUSEHOLDS TRADE IN GOODS JAPANESE HOUSEHOLDS FINANCIERS PROFITS PROFITS Two important papers in 1976: Dornbusch s overshooting model, and Kouri s portfolio balance model Obstfeld, Rogoff (1995) brought Mundell-Fleming-Dornbusch model in modern macroeconomics We provide a modern general equilibrium theory of the financial market forces first sketched by Kouri

5 Related Open Macroeconomics Literature Portfolio Flows: Hau, Rey (2006); Bacchetta, Van Wincoop (2010); Jeanne and Rose (2002); Evans, Lyons (2002) Financial Intermediation: Maggiori (2011); Bruno and Shin (2012) Welfare and Policy Analysis: Farhi, Werning (2012a,b); Farhi, Gopinath, Itskhoki (2012); Magud, Reinhart, Rogoff (2012); Devereux, Engel (2003); Schmitt-Grohe, Uribe (2012a,b); Korinek (2011); Bianchi (2011) Complete Market Asset Pricing: Lucas (1982); Backus, Kehoe, Kydland (1992); Dumas (1992); Obstfeld, Rogoff (2001); Pavlova, Rigobon (2007); Verdelhan (2010); Farhi, Gabaix (2011); Hassan (2013); Colacito, Croce (2011) Incomplete Market Asset Pricing: Alvarez, Atkeson, Kehoe (2002,09)

6 Basic Model We present here the simplest model: real model, imperfect capital markets Two countries (US, Japan (*)). Two periods (t = 0, 1) Unit measure of households in each country Four goods: 1 non-tradable (NT) and 1 tradable good in each country NT are endowments, tradables produced with int. immobile inelastically supplied labor NT good is the numéraire in each economy Incomplete Markets: two risk-free bonds that pay for sure one unit of the domestic numéraire (the NT good) for each economy Households borrow/lend in domestic risk-free bonds with the financiers Financiers intermediate resulting global capital flows

7 The Household Problem US households consumption/saving decision: max c s.t. E [θ 0 ln C 0 + βθ 1 ln C 1 ] 1 t=0 C NT,t + p H,t C H,t + p F,t C F,t R t 1 t=0 Y NT,t + p H,t Y H,t R t where C t [(C NT,t ) χt (C H,t ) at (C F,t ) ιt ] 1 θt, and θ t = χ t + a t + ι t

8 The Household Problem US households consumption/saving decision: max c s.t. E [θ 0 ln C 0 + βθ 1 ln C 1 ] 1 t=0 C NT,t + p H,t C H,t + p F,t C F,t R t 1 t=0 Y NT,t + p H,t Y H,t R t where C t [(C NT,t ) χt (C H,t ) at (C F,t ) ιt ] 1 θt, and θ t = χ t + a t + ι t Corresponding Japanese households problem: where C t max c E [θ0 ln C0 + β θ1 ln C1 ] s.t. 1 CNT,t + p H,t C H,t + p F,t C F,t R t t=0 1 YNT,t + p F,t Y F,t + π t t=0 R t [ ] 1 (CNT,t )χ t (C H,t ) ξt (CF θ,t )a t t ; and θt = χ t + at + ξ t

9 Household Optimality Conditions Intra-temporal Optimality Conditions: C NT,t = χ t λ t ; p F,t C F,t = ι t λ t Simplifying assumption: Y NT = χ t λ t = 1. US imports: pf,t C F,t = ι t ; Japan Imports: p H,t C H,t = ξ t US net exports: NXt = ξ t e t ι t e t : (real) Dollar per Yen exchange rate

10 Household Optimality Conditions Intra-temporal Optimality Conditions: C NT,t = χ t λ t ; p F,t C F,t = ι t λ t Simplifying assumption: Y NT = χ t λ t = 1. US imports: pf,t C F,t = ι t ; Japan Imports: p H,t C H,t = ξ t US net exports: NXt = ξ t e t ι t e t : (real) Dollar per Yen exchange rate Inter-temporal Optimality Conditions: [ ] 1 = E βr U 1,C NT U 0,C NT [ = E βr χ ] 1/C NT,1 = βr, χ 0 /C NT,0

11 Financiers Asset Demand Unit measure of intermediaries, each financier runs one intermediary Agents are selected at random. Zero starting capital. Rebate all profits to households Trade Dollar and Yen bonds. Balance sheet: q 0 = q F,0 e 0 Financiers maximize expected returns in dollars: [ ( V 0 = E β R R e )] 1 q 0 e 0 Intermediation Friction: After taking positions, but before uncertainty is realized ( financiers ) can divert funds. If financiers divert, creditors recover 1 Γ q 0 e 0 of their claims q 0 : e 0 ( ) V 0 q 0 e }{{} 0 e 0 Γ q 0 2 q0 e 0 = Γ e 0 }{{}}{{}}{{} Intermediary Value in Yen Total Claims Diverted Portion Total divertable Funds

12 Financiers Asset Demand: Micro-foundations Financiers problem: [ ( )] max q0 V 0 = E β R R e 1 e 0 q 0 s.t. V 0 Γ q2 0 e 0 Optimality Constraint always binds Financiers demand q 0 dollar and q 0 /e 0 yen, according to: q 0 = 1 [ ] Γ E e 0 R R e 1 Γ : no amount of intermediation is possible autarky Γ = 0: any amount of intermediation is possible Uncovered Interest Parity holds This Γ demand function is key to the model: Basic Gamma model Simplifying assumption: financiers pay all profits to Japanese households

13 Financiers Asset Demand: Micro-foundations Financiers problem: [ ( )] max q0 V 0 = E β R R e 1 e 0 q 0 s.t. V 0 Γ q2 0 e 0 Optimality Constraint always binds Financiers demand q 0 dollar and q 0 /e 0 yen, according to: Q 0 = 1 [ ] Γ E e 0 R R e 1 Γ : no amount of intermediation is possible autarky Γ = 0: any amount of intermediation is possible Uncovered Interest Parity holds This Γ demand function is key to the model: Basic Gamma model Simplifying assumption: financiers pay all profits to Japanese households

14 Equilibrium Exchange Rate The flow equations in the US dollar market: ξ 0 e 0 ι 0 + Q 0 = 0; ξ 1 e 1 ι 1 RQ 0 = 0; Q 0 = 1 [ ] Γ E e 0 R R e 1 The equilibrium exchange rate follows (assume ξ t = R = R = 1): e 0 = (1 + Γ) ι [ ] 0 + E[ι 1 ] e0 e 1 ; E = Γ (E [ι 1] ι 0 ) 2 + Γ (1 + Γ) ι 0 + E[ι 1 ] e 0 Financial Autarky (Γ ): e t = ι t UIP (Γ 0): e 0 = E[e 1 ] = ι 0+E[ι 1 ] 2

15 Overview of Results 1. Exchange Rate Disconnect. Some Empirical Evidence Little connection between traditional fundamentals and exchange rates Meese, Rogoff (1983) More evidence that exchange rates are connected to flows in the medium run Adrian, Etula, Groen (2011), Adrian, Etula, Shin (2013): financiers balance sheet forecast USD FX Hau, Massa, Peress (2010): inflows cause currency appreciation. Clean IV approach Yogo, Hong (2012): CME speculators positions help predict currency returns Froot, Ramodorai (2005): flows are associated with most of the variation in expected currency returns over medium horizons, fundamentals matter only at long horizon

16 Overview of Results 1. Exchange Rate Disconnect. Some Empirical Evidence 2. Gross Capital Flows Matter For simplicity, assume that some Japanese households have a noise demand f for Dollar bonds (financed in Yen bonds), then the equilibrium exchange rate follows: e 0 = (1 + Γ) ι 0 + E[ι 1 ] f Γ 2 + Γ e0 f = Γ 2+Γ : if Japanese households demand Dollar bonds (f > 0), then the Dollar appreciates ( e 0 ): supply and demand of assets matters! This effect is absent both in complete market models or in models that assume UIP. Empirical support: Hau et al. (2010)

17 Overview of Results 1. Exchange Rate Disconnect. Some Empirical Evidence 2. Gross Capital Flows Matter 3. Flows not just Stocks Matter US has an exogenous Dollar-denominated debt toward Japan. D0 due at time zero, and D 1 due at time one e 0 = (1 + Γ) ι 0 + E [ι 1 ] 2 + Γ When finance is imperfect (Γ > 0): + (1 + Γ) D 0 + D Γ Timing of repayment (flow, D 0 D 1 ) matters, not just debt stock (D 0 + D 1) The higher Γ the more weight on early repayment

18 Overview of Results 1. Exchange Rate Disconnect. Some Empirical Evidence 2. Gross Capital Flows Matter 3. Flows not just Stocks Matter 4. External Debt and Currency Returns US net foreign assets: N 0 + = e 0 ι 0 = E[ι1] ι0 2+Γ Proposition: When there is a financial disruption ( Γ), countries that are net external debtors (N 0 + < 0) experience a currency depreciation ( e), while the opposite is true for net-creditor countries Suppose ι 0 E[ι 1] > 0, US runs a trade deficit and borrows in dollars Financiers are long Dollar and short Yen (Q 0 > 0) If financial conditions worsen ( Γ), the Dollar depreciates ( e 0) Empirical support: Della Corte, Riddiough and Sarno (2013)

19 Overview of Results 1. Exchange Rate Disconnect. Some Empirical Evidence 2. Gross Capital Flows Matter 3. Flows not just Stocks Matter 4. External Debt and Currency Returns 5. Carry Trade Model extension to 3 periods with Γ1 stochastic Carry trade return: R c 1 R e1 e 0 1, with R R R Expected carry trade returns: E[R c 1 ] = Γ 1 1+Γ 1 E 0 [ Γ1 1+Γ 1 ]. Returns are higher: R Γ 0 (Γ R ) Γ 1 (Γ 0 + R ) + Γ 0 + (R ) 2 the higher is the interest rate differential R the worse financial conditions are today ( Γ 0), the better they are tomorrow ( {Γ 1, Γ 1}) CIP always holds, separate class of financiers does all arbitrages

20 Overview of Results 1. Exchange Rate Disconnect. Some Empirical Evidence 2. Gross Capital Flows Matter 3. Flows not just Stocks Matter 4. External Debt and Currency Returns 5. Carry Trade 6. Welfare and Heterodox Financial Policies

21 Output Effects of Exchange Rates Output: Y = L; potential labor (L) is supplied inelastically Assume downwards rigid goods prices at p H, and producer currency pricing (PCP), then: C H,t + CH,t = a ( ) t + e t ξ t at + e t ξ t Y H,t = min, L p H p H Output can be demand determined Intuitively: if Dollar prices are too high ( p H ), or if the Dollar is too strong ( e t ), output falls and there is unemployment (Y H < L) Recall e t = f (Γ t, Q H,t ): so financial determination of ER can have real effects

22 FX Swaps and Currency Interventions: Welfare Consequences ( ) Recall: Y H,0 (e 0 ) = min a0 +e 0 ξ 0 p, L H,0 The government buys q yen and sells qe o dollars at time 0 So e 0 (q) = 1 + Γ 2+Γ q + O(q2 ): Dollar depreciates, creating employment Welfare: V (q) E[U 0 + U 1 ] = V FB + a 0 ln Y H,0(e 0 (q)) L + O(q 2 ) Proposition If Γ > 0 and Y H,0 (q = 0) < L, then welfare V (q) is increasing in intervention q [0, q Opt ], where e(q Opt ) generates full employment Note that under no taste shocks there are no private incentives to intervene

23 Take-Aways We presented a basic model with: Imperfect capital markets: financial intermediation, supply and demand of assets matters! Production: real effects of ER fluctuations, unemployment (Potentially) sticky prices: PCP, LCP, incomplete pass-through Welfare analysis: monetary policy, heterodox financial policies Key implications: Exchange rates are a financial phenomenon determined by supply and demand of assets in different currencies. Financiers balance sheets and risk tolerance are important determinants of ER Key take-away: Floating exchange rate regimes can be the source of problems. Heterodox policies (interventions, capital controls) make sense when imbalances are big and financial markets distressed

24 Coming Soon The framework is quite flexible and can also accommodate: Trading in multiple assets Multiple countries Quantitative implications (numerically) Testing the key empirical implications: work in progress Model is consistent with a number of facts: carry trade, debtor and creditor countries currency risk, failure of PPP, exchange rate disconnect, Backus and Smith condition Directly test channel Financiers demand equation: relates balance sheet and risk tolerance to currency returns

25 FX and Bond Market Structure This paper is not about volume, turnover, and transaction costs: $5.3trn daily volume in FX market Turnover mostly associated with market-making and inventory trading (Lyons (1997)) We abstracted away these microstructure effects via perfect-risk-sharing among financiers This paper is about risk bearing capacity of key financial players over several months/quarters: International FX and bond markets are OTC role for intermediation Relatively concentrated market in ultimate risk taking PIMCO, Blackrock, Goldman Sachs (GSAM), Soros take most of the risk are the marginal player While large, these institutions have limited risk taking capacity