International Monetary Policy Transmission through Banks in Small Open Economies. S. Auer, C. Friedrich, M. Ganarin, T. Paligorova, P.

Transcription

1 International Monetary Policy Transmission through Banks in Small Open Economies S. Auer, C. Friedrich, M. Ganarin, T. Paligorova, P. Towbin

2 Disclaimer The views expressed in this paper are our own and do not necessarily represent those of the Bank of Canada or those of the Swiss National Bank.

3 Introduction

4 Motivation Using the examples of Canada and Switzerland, this paper studies the international transmission of monetary policy through banks in small open economies Small open economies are substantially affected by international economic conditions but have little effect on international economic conditions themselves There is a large literature that documents that small open economies are affected by monetary policy through real channels (e.g., demand or exchange rate effects) However, whether and how monetary policy transmits internationally through banks and in the context of small open economies is an open question

5 Comparison of Banking Systems The Canadian and the Swiss banking system have several features in common: Both countries have a large neighbour economy Foreign banks do not have a relevant market share Domestic lending was not negatively affected by the global financial crisis There are, however, some notable differences that shape our empirical approach: While the Canadian banking system is highly concentrated, the Swiss banking system features a substantially larger number of banks The inward transmission approach is estimated more precisely if the number of domestic banks is large While most banks headquartered in Switzerland are not very active in lending abroad, Canadian banks have a significant share of their lending operations abroad The outward transmission approach has a higher external validity if domestic banks are active abroad

6 Setup of the Paper Domestic Monetary Policy Foreign Monetary Policy

7 Setup of the Paper Outward Transmission for Canada Domestic Monetary Policy Foreign Monetary Policy

8 Setup of the Paper Inward Transmission for Switzerland Domestic Monetary Policy Foreign Monetary Policy

9 The IBRN Project focuses on two channels of international monetary policy transmission. Bank lending channel (e.g., Bernanke and Blinder, 1988; Kashyap and Stein, 1994) Impact on domestic and foreign lending goes in the same direction Tighter domestic or foreign MP raises banks' funding costs => Banks reduce domestic and foreign lending Portfolio channel (e.g., Bernanke and Gertler, 1995, balance sheet channel ) Opposite impact on domestic and foreign lending When domestic MP tightens, domestic borrowers become more risky => Banks substitute away from domestic lending to safer foreign lending Equivalently, a tightening of foreign MP triggers a shift to safer domestic lending

10 Outward Transmission of Domestic Monetary Policy (Canada)

11 Empirical Specification K K = 0 Channel b Y b, j, t X b, t 1 f j, t f b MP t dom k, j, t K 1 lending growth of Canadian banks foreign affiliates change in Canadian monetary policy (in t to t-3) variable to test for transmission channels (in t-4) bank controls (e.g., asset size, capital, liquidity) country-time fixed effects bank fixed effects

12 Data Bank-level data: The Canadian banking system is highly concentrated and largely dominated by six domestically owned banks; of which we include four in the empirical analysis Dependent variable: Lending growth of Canadian banks foreign affiliates 31 destination countries Sample period: 2000Q1-2015Q4 Measure of monetary policy: Change in the Canadian monetary policy interest rate Taylor residual based on a regression of the forecast error (FCE) in the 3-month interest rate on FCEs of inflation and output growth

13 Transmission Variables International Bank Lending Channel Prediction for a domestic MP tightening: banks decrease foreign lending Factors that can mitigate this effect (channel variables): Less dependence on domestic funding (in particular, short-term deposit funding) More collateral that allows to access cheap wholesale funding (e.g., more Tier 1 capital) International Portfolio Channel Prediction for a domestic MP tightening: banks increase foreign lending Factors that can mitigate this effect (channel variables): Higher exposure to non-affected foreign borrowers (e.g., many claims on foreign borrowers)

14 Results Evidence of Channels Channel Bank Lending Channel (1a) (1a) (2a) (2b) (3a) (3b) Short Term (Deposit) Funding Ratio Tier 1 Ratio Portfolio Channel Total Total Claims on Claims on Foreign Foreign Borrowers Borrowers / Total / Total Assets Assets ΔRate Taylor Res. Mon. Pol. Measure ΔRate Taylor ΔRate Taylor Res. Res. Expected Sign neg. neg. pos. pos. neg. neg. ΣΔMP Domestic_t to t-3(*channel_t-4) ** ** ** [0.030] [0.921] [0.031] [0.160] [0.304] [0.035] (Impact) ΔMP Domestic_t*Channel_t * ** [0.075] [0.485] [0.827] [0.049] [0.770] [0.823] Channel_t * [0.143] [0.089] [0.097] [0.695] [0.606] [0.593] Country-Time fixed effects Yes Yes Yes Yes Yes Yes Bank fixed effects Yes Yes Yes Yes Yes Yes Observations ,975 4,975 4,975 4,975 4,975 R-squared Adjusted R-squared Number of banks Number of countries Evidence of both an international bank lending and an international portfolio channel in Canada Effects are sizable: Coefficient in Spec. (1a): Δ in Canadian policy rate: 100bp Difference in local lending by Canadian banks foreign affiliates - 25th vs. 75th percentile - after a policy change: percent In Canada, the two channels seem to balance each other (aggregate specification turns out insignificant)

15 Inward Transmission of Foreign Monetary Policy (Switzerland)

16 Specification 1: What is the effect of foreign mp on domestic lending growth? 4 Y b,t = α 0 + α cccc 1,k MM cccc t k + α cccc 2,k QQ cccc t k ZZZ cccc t k=0 We estimate the effect of foreign mp in three major currency areas separately (US, euro area, UK). Horizon: four quarters. Conventional for monetary policy ( MM cccc t k ): change in policy rate cccc Unconventional monetary policy( QQ t k ): change in CB Assets/GDP Controls: bank balance characteristics X b,t 1 ; domestic/foreign cyclical conditions and domestic mp (Z t 1 ). We focus on domestic banks (two G-SIBs and domestic retail banks). + α cccc 3 ZZZ cccc t + + α 4 Z t 1 + α 5 X b,t 1 + f b +ε b,t, Data sources: IBRN monetary policy, SNB banking statistics, 2000Q1-2015Q4

17 Contractionary foreign mp either has a small positive or no significant effect on domestic lending. No channel: MP US No channel: MP EA No channel: MP GB ΣΔMP country i_t to t * 0.003* [0.117] [0.061] [0.084] ΣΔQE country i_t to t-3*zlb ** [0.792] [0.011] [0.651] [0.001] [0.000] [0.000] Observations R-squared Number of banks Results provide preliminary evidence on importance of transmission channels: o o Not consistent with a dominant role of international bank lending channel (sign should be negative) Consistent with: i) a role for portfolio channel only or ii) with a combination of both channels. To draw further inferences, we need to look at variation across banks.

18 Specification 2: How does the effect of foreign MP vary across banks? 4 Y b,t = α 0 + α cccc 1,k MM cccc t k + α cccc 2,k QQ cccc t k ZZZ cccc t k=0 Focus is on variation across banks (ccccccc cccc b,t 4 ). We control for cyclical conditions with time fixed effects. International portfolio channel (asset side): When foreign monetary policy contracts, bank with more foreign claims should have higher domestic lending growth (portfolio reallocation). International bank lending channel (liability side): ccccccc cccc b,t 4 +α cccc 3 ZZZ cccc t ccccccc cccc b,t f t + ε b,t, When foreign monetary policy contracts, bank with more foreign funding should have lower domestic lending growth (higher funding costs) Data: BIS Locational Banking Statistics to measure bilateral foreign claims and foreign funding

19 Variation across banks is mostly consistent with a role for the portfolio channel Channel: (MP US) Cross- Border Assets in US / T otal Assets (MP EA) Cross- Border Assets in EA / T otal Assets ΣΔMP country i_t to t-3(*channel_t-4) 0.196** ** [0.037] [0.346] [0.011] ΣΔQE country i_t to t-3(*channel_t-4)*zlb 0.101* (MP GB) Cross- Border Assets in GB / T otal Assets Observations R-squared Banks with more foreign claims tend to have higher lending growth (in 2 out of 3 specifications). Consistent with a higher risk of foreign borrowers and a reallocation of lending.

20 No evidence with role for international bank lending channel. Channel: (MP US) Crossborder Liabilities from US / Total Assets (MP EA) Crossborder Liabilities from EA / Total Assets ΣΔMP country i_t to t-3(*channel_t-4) 0.210*** ** [0.000] [0.440] [0.030] ΣΔQE country i_t to t-3(*channel_t-4)*zlb [0.221] [0.605] [0.850] (MP GB) Crossborder Liabilities from GB / Total Assets Observations R-squared Banks with more foreign funding tend to have higher lending growth, opposite from what bank lending channel predicts. Bank lending channel theory may not be valid for all liabilities, in particular bank liabilities and wholesale funding in general (Drechsler et al. 2016).

21 Distinguishing between cross border funding from banks and non-banks matters. Channel: (2a-2) (2b-2) (2c-2) (2a-3) (2b-3) (2c-3) (MP US) Crossborder Bank Liabilities from US / T otal Assets (MP EA) Crossborder Bank Liabilities from EA / T otal Assets (MP GB) Crossborder Bank Liabilities from GB / T otal Assets (MP US) Crossborder Non-bank Liabilities from US / T otal Assets (MP EA) Crossborder Non-bank Liabilities from EA / T otal Assets ΣΔMP country i_t to t-3(*channel_t-4) 0.204*** *** *** [0.001] [0.817] [0.001] [0.387] [0.203] [0.006] ΣΔQE country i_t to t-3(*channel_t-4)*zlb *** [0.333] [0.923] [0.434] [0.170] [0.549] [0.000] (MP GB) Crossborder Non-bank Liabilities from GB / T otal Assets Observations R-squared Positive coefficient is driven by liabilities to foreign banks Consistent with an indirect portfolio channel: Foreign banks shift their money abroad. Available funds for domestic banks increase => increase in domestic lending.

22 Conclusion

23 Conclusion Inward transmission of foreign monetary policy through banks in Switzerland: Consistent with a role for the international portfolio channel. No evidence for the traditional international bank lending channel. Outward transmission of domestic monetary policy through banks in Canada: Both channels work as predicted. They largely balance each other. Overall, for both, outward and inward transmission, the aggregate effect on bank lending appears to be limited. Policy implication: despite strong real linkages, international transmission of monetary policies in small open economies through banks can be limited, depending on funding and balance sheet structure.