Dollar Funding of Global banks and Regulatory Reforms: Evidence from the Impact of Monetary Policy Divergence

Size: px

Start display at page:

Download "Dollar Funding of Global banks and Regulatory Reforms: Evidence from the Impact of Monetary Policy Divergence"

Virginia Skinner
5 years ago
Views:

Dollar Funding of Global banks and Regulatory Reforms: Evidence from the Impact of Monetary Policy Divergence Nao Sudo Monetary Affairs Department Bank of Japan Prepared for

1 Dollar Funding of Global banks and Regulatory Reforms: Evidence from the Impact of Monetary Policy Divergence Nao Sudo Monetary Affairs Department Bank of Japan Prepared for Symposium: CIP-RIP? at Bank for International Settlements, May 2017 The views expressed in this presentation are those of the author and do not necessarily reflect those of the BOJ.

2 Outline of paper (1) Why do we care about CIP deviations? (2) What we do. Construct a simple model with U.S. and JPN banks extending the model of Ivashina et al. (2015). Derive two predictions from the model. 1. Determinants of CIP deviations. 2. Effects of regulatory reforms. Check if the model accords with the data, using four currency pairs (CHF, EUR, GBP, JPY with USD) 2

3 Motivation Why do we care about CIP deviations? 3

4 Reason 1: Historically, CIP deviations and banking crisis often came together. CIP deviation Expected default frequency of banks 1.6 % 7 % % EDF of Japanese banks USD/JPY EUR/USD GBP/USD EDF of Eurozone banks EDF of UK banks Japan premium (rhs) CY CY Notes: 1. The shadow areas correspond to the period of the Japan s financial crisis (November 1997 through January 1999), the global financial crisis (December 2007 through June 2009), and the Eurozone sovereign crisis (May 2011 through June 2012). 2. A CIP deviation is computed from FX swap-implied dollar rates and Libor rates. 3. As regards banks default probability, an average of the EDF (Expected Default Frequency) of all GSIBs that are headquartered in each jurisdiction is shown. Japanese Premium is calculated as 3-month USD TIBOR rate minus 3-month USD LIBOR rate. Sources: Bloomberg; Moody s; BOJ 4

5 Reason 2: CIP deviation is a measure of funding liquidity risks associated with global banks crossborder lending. USD-denominated foreign positions trill. USD Non-US banks USD-denominated foreign claims (A) Non-US banks USD-denominated foreign liabilities (B) Implied cross currency funding (ie FX swaps) % Non-US banks implied crosscurrency funding ratio (A)-(B) (A) Global Financial Crisis Eurozone Sovereign Crisis 0 CY Sources: BIS consolidated statistics (immediate borrower basis); BIS locational statistics 4 CY

6 Reason 3: CIP deviations nowadays reflect reactions of global banks to a change in cross-border investments due to a change in monetary policy. 200 bill. EUR Outward portfolio investment Euro area Investment in the other countries Investment in the U.S. Portfolio Investment 15 trill. JPY Japan Investment in the other countries Investment in the U.S. Portfolio Investment FRB tapering ECB monetary easing -100 CY Sources: ECB, BOJ, Balance of Payments FRB rate hike 0-5 FRB tapering FRB rate hike -10 QQE with YCC BOJ monetary easing -15 CY

7 Model Settings 7

8 Simple equilibrium model of FX swap market Flow of USD Flow of JPY ( ) US banks and other arbitrageurs ( ) ( + ) ( ) ( + ) USD bond and loan markets FED USD funding and deposit markets FX swap market JPY funding and deposit markets BOJ JPY bond and loan markets ( + ) ( ) ( + ) Japanese banks (Non-US financial institutions) ( ) (q) Notes: 1. Figures in parentheses are interest rates. 2. is CIP deviation. and are risk-free rates. and are risk premiums. 8

9 Simple equilibrium model of FX swap market Flow of USD Flow of JPY ( ) US banks and other arbitrageurs ( ) ( + ) ( ) ( + ) USD bond and loan markets FED USD funding and deposit markets FX swap market JPY funding and deposit markets BOJ JPY bond and loan markets Return from USD denominated loan ( + ) ( ) ( + ) Japanese banks (Non-US financial institutions) ( ) (q) Notes: 1. Figures in parentheses are interest rates. 2. is CIP deviation. and are risk-free rates. and are risk premiums. 9

10 Simple equilibrium model of FX swap market Flow of USD Flow of JPY ( ) US banks and other arbitrageurs ( ) ( + ) ( ) ( + ) USD bond and loan markets FED USD funding and deposit markets FX swap market JPY funding and deposit markets BOJ JPY bond and loan markets Funding rate of USD from US money market ( ) ( + ) ( + ) Japanese banks (Non-US financial institutions) ( ) (q) Notes: 1. Figures in parentheses are interest rates. 2. is CIP deviation. and are risk-free rates. and are risk premiums. 10

11 Simple equilibrium model of FX swap market Flow of USD Flow of JPY ( ) US banks and other arbitrageurs ( ) ( + ) ( ) ( + ) USD bond and loan markets FED USD funding and deposit markets FX swap market JPY funding and deposit markets BOJ JPY bond and loan markets ( + ) Borrowing cost of USD from FX swap ( ) ( + ) Japanese banks (Non-US financial institutions) ( ) (q) Notes: 1. Figures in parentheses are interest rates. 2. is CIP deviation. and are risk-free rates. and are risk premiums. 11

12 Simple equilibrium model of FX swap market Flow of USD Flow of JPY ( ) US banks and other arbitrageurs ( ) ( + ) ( ) ( + ) Return from JPY denominated loan USD bond and loan markets FED USD funding and deposit markets FX swap market JPY funding and deposit markets BOJ JPY bond and loan markets ( + ) ( ) ( + ) Japanese banks (Non-US financial institutions) ( ) (q) Notes: 1. Figures in parentheses are interest rates. 2. is CIP deviation. and are risk-free rates. and are risk premiums. 12

13 Optimization problem of Japanese banks max $, \, $, \, $ + \ $ \ + Subject to where $ + = $ + \ = \, = 1+ 2, = 1+ 2, = , =

14 Optimization problem of Japanese banks max $, \, $, \, $ + \ $ \ + Subject to where $ $ + = $ + \ = \, = + $ $, : USD denominated loan Japanese banks increase when interest rate is high. Lending becomes costly when is high. = 1+ 2, = , =

15 Optimization problem of Japanese banks max $, \, $, \, $ + \ $ \ + Subject to where $ + = $ + \ = \, = 1+ 2, \ = + \ \, JPY denominated loan Japanese banks increase when interest rate q is high. Lending becomes costly when is high. = , =

16 Optimization problem of Japanese banks max $, \, $, \, $ + \ $ \ + Subject to where $ $ + = $ + \ = \, = 1+ 2, = 1+ 2, = + + $ + $, : USD funding from money markets Funding rate increases with banks default probability. Funding is costly when high. is = : risk-free rate in USD : premium 16

17 Optimization problem of Japanese banks max $, \, $, \, $ + \ $ \ + Subject to where $ + = $ + \ = \, = 1+ 2, : risk-free rate in JPY = 1+ 2 = , \ = + \ + \,. : JPY funding from money markets in Japan Funding is costly when high. is 17

18 Optimization problem of Japanese banks max $, \, $, \, $ + \ $ \ + S: Funding of USD from FX swap Subject to where $ + = $ + \ = \, = 1+ 2, : CIP deviation = 1+ 2, = , =

19 Optimization problem of Japanese banks max $, \, $, \, $ + \ $ \ + M: Cash holding in USD due to precautionary motives Subject to where $ + = $ + \ = \, = 1+ 2, (1) M yields zero return (2) V captures degree of uncertainty = 1+ 2, = , =

20 Optimization problem of Japanese banks max $, \, $, \,, $ + \ $ \ + Subject to where $ + = $ + \ = \, = 1+ 2, Upper equality: BS of USD Lower equality: BS of JPY = 1+ 2, = , =

21 Optimization problem of US banks max $, \, $, \,, h $ +h \ $ + \ Subject to where $ + = + $ + \ = \, h $ h \ $ = 1+ $ 2 $, = 1+ \ 2 \, = 1+ $ + 2 $, \ = 1+ + \ + 2 \. 21

22 Optimization problem of US banks Subject to where max $, \, $, \,, h $ h \ h $ +h \ $ + $ + = + $ + \ = \, = 1+ $ 2 $, = 1+ \ 2 \, \ : USD provided by Real money investors to FX markets The amount of is exogenously given. $ \ = 1+ $ + 2 $, = 1+ + \ + 2 \. 22

23 Model Predictions 23

24 Determinants of CIP deviations: model prediction = CIP deviation Δ (Price of FX swap) Prediction I: A widening in interest margin differential (IMD) increases since it is more profitable for Japanese banks to invest in USD denominated assets, and less profitable for US banks to invest in JPY denominated assets. IMD, * V*, W* IMD,, V S (Size of FX swap position) 24

25 Determinants of CIP deviations: model prediction = Prediction II: A higher default probability of Japanese banks increases since it is more costly for Japanese banks to fund USD from US money market. CIP deviation Δ (Price of FX swap) IMD, * V*, W* IMD,, V S (Size of FX swap position) 25

26 Determinants of CIP deviations: model prediction = Prediction III: A higher default probability of US banks decreases since it is more costly for US banks to fund JPY from Japanese money market. CIP deviation Δ (Price of FX swap) IMD, * V*, W* IMD,, V S (Size of FX swap position) 26

27 Determinants of CIP deviations: model prediction = Prediction IV: A larger uncertainty increases since it increases precautionary demand for USD of both US and Japanese banks. CIP deviation Δ (Price of FX swap) IMD, * V*, W* IMD,, V S (Size of FX swap position) 27

28 Determinants of CIP deviations: model prediction = Prediction V: A decline in endowment of real money investors increases since it decreases USD supplied in FX markets. CIP deviation Δ (Price of FX swap) IMD, * V*, W* IMD,, V S (Size of FX swap position) 28

29 Effects of Regulatory reforms: model prediction Cost function of USD funding Non-US banks: USD suppliers: + + $ + $ + $ + $ Regulatory reform makes USD funding and/or expanding the size of BS more costly for banks. 29

30 Effects of Regulatory reforms: model prediction Prediction: becomes more responsive to a change in economic environments when or is higher Ex. = >0, = >0. CIP deviation Δ Strict regulation (high and ) IMD IMD Size of FX swap position CIP deviation Δ Loose regulation (low and ) IMD IMD Size of FX swap position 30

31 Empirical results 31

32 Determinants of CIP deviations: Panel regressions of CIP deviations (3M) in four currency pairs, EUR/USD, USD/JPY, GBP/USD, and USD/CHF Model 1 Model 2 EUR/USD USD/JPY USD/CHF GBP/USD Policy divergence Interest margin differential 0.06 *** 0.06 *** 0.07 *** 0.04 ** 0.10 *** Default probability of banks EDF of US banks EDF of non-us banks *** ** *** *** 0.09 *** 0.12 *** *** 0.13 *** VIX Proxy for precautionary demand for USD Fixed effects 0.01 *** 0.02 *** 0.01 * 0.01 *** 0.01 *** Yes Yes R-squared RMSE No. of observations Notes: 1. Sample period: 2007M1-2016M ***, **, and * respectively indicate significance at the 1%, 5%, and 10% level. 32

33 Determinants of CIP deviations: Panel regressions of CIP deviations (3M) in four currency pairs, EUR/USD, USD/JPY, GBP/USD, and USD/CHF Model 1 Model 2 EUR/USD USD/JPY USD/CHF GBP/USD Policy divergence Default probability of banks Interest margin differential EDF of US banks EDF of non-us banks VIX Fixed effects Proxy for precautionary demand for USD 0.06 *** 0.06 *** 0.07 *** 0.04 ** 0.10 *** *** ** *** *** 0.09 *** 0.12 *** *** 0.13 *** 0.01 *** 0.02 *** 0.01 * 0.01 *** 0.01 *** Yes Yes R-squared RMSE No. of observations Notes: 1. Sample period: 2007M1-2016M ***, **, and * respectively indicate significance at the 1%, 5%, and 10% level. 33

34 Determinants of CIP deviations: Panel regressions of CIP deviations (3M) in four currency pairs, EUR/USD, USD/JPY, GBP/USD, and USD/CHF Model 1 Model 2 EUR/USD USD/JPY USD/CHF GBP/USD Policy divergence Default probability of banks Interest margin differential EDF of US banks EDF of non-us banks VIX Proxy for precautionary demand for USD Fixed effects 0.06 *** 0.06 *** 0.07 *** 0.04 ** 0.10 *** *** ** *** *** 0.09 *** 0.12 *** *** 0.13 *** 0.01 *** 0.02 *** 0.01 * 0.01 *** 0.01 *** Yes Yes R-squared RMSE No. of observations Notes: 1. Sample period: 2007M1-2016M ***, **, and * respectively indicate significance at the 1%, 5%, and 10% level. 34

35 Determinants of CIP deviations: Role of MP divergence (%, Deviation from average) Decomposition of CIP deviation (3M) EUR/USD Central banks' balance sheet EDF VIX Residual CIP deviation (%, Deviation from average) USD/JPY CY Central banks' balance sheet -0.4 EDF VIX -0.6 Residual CIP deviation -0.8 CY (Note) Central banks balance sheet in each panel is the contribution of the difference in growth rates of the central banks balance sheets. EDF in each panel is the sum of the contribution of EDF in the non-u.s. jurisdiction and EDF in the U.S., indicating the net effect of banks default probabilities. Residuals in each panel include the contribution of control variables. 35

36 Determinants of CIP deviations: Further decomposing CIP deviations to extract (%, Deviation from average) Decomposition of CIP deviation (3M) EUR/USD Central banks' balance sheet EDF VIX Residual CIP deviation Residual contains information about. USD/JPY We extract by imposing a sign (%, Deviation from average) restriction using data of both S and CY Central banks' balance sheet -0.4 EDF VIX -0.6 Residual CIP deviation -0.8 CY (Note) Central banks balance sheet in each panel is the contribution of the difference in growth rates of the central banks balance sheets. EDF in each panel is the sum of the contribution of EDF in the non-u.s. jurisdiction and EDF in the U.S., indicating the net effect of banks default probabilities. Residuals in each panel include the contribution of control variables. 36

37 Oil price (y/y % chg.) Determinants of CIP deviations: Nature of and oil price Relationship between oil price and identified W [ ] [ ] Contributions of W^ on CIP deviation Oil price (y/y % chg.) [ 0.30 ] [ 0.69 ] Contributions of W^ on FX swap transaction volume S Note: Figures in parentheses indicate correlation coefficients. 37

38 Determinants of CIP deviations: EM economy-cip deviation linkages through US monetary policy (Interest-rate hike)? spillover Slowdown in EM economies Fall in oil prices Capital outflows from EM While arbitrage trading by banks have declined due to regulations, real money investors have enhanced their presence in the FX swap market Δ $ non-us banks curtail the USD-denominated lending to EM 38

39 Effects of Regulatory reforms = + +. Sample period: Jan Dec. 2016, < =, 39

40 Effects of Regulatory reforms = + +. Sample period: Jan Dec , < =, CY T CY T Note: The X-axis denotes the starting period from which the dummy variable takes unity. Dotted lines indicate 95 percent confidence intervals. 40

41 Effects of Regulatory reforms = + +. Sample period: Jan Dec , < =, Increase in IMD results in about times larger response of 0.25 CIP deviation in 2016 than 2012! CY T CY T Note: The X-axis denotes the starting period from which the dummy variable takes unity. Dotted lines indicate 95 percent confidence intervals. 41

42 Summary Banks creditworthiness is being replaced by other factors as the driving force of CIP deviations, though it is hasty to conclude CIP deviations should be dropped from watch list. New characters in the play: disagreement of monetary policy stance and real money investors. Volatile CIP deviations as byproduct of safeguarding financial system? 42

43 Thank you 43

Monetary Policy Divergence and Global Financial Stability: From the Perspective of Demand and Supply of Safe Assets

Monetary Policy Divergence and Global Financial Stability: From the Perspective of Demand and Supply of Safe Assets January, 7 Speech at a Meeting Hosted by the International Bankers Association of Japan