A Policy Model for Analyzing Macroprudential and Monetary Policies

Transcription

1 A Policy Model for Analyzing Macroprudential and Monetary Policies Sami Alpanda Gino Cateau Cesaire Meh Bank of Canada October 2014 Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

2 Disclaimer The views expressed in this paper are those of the authors. No responsibility should be attributed to the Bank of Canada. Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

3 Motivation Recent global financial crisis was a reminder that the real economy and the financial system are closely linked financial system can be a source of shocks amplify/propagate shocks originating elsewhere Theoretical front: renewed interest in incorporating real-financial linkages into DSGE models Policy front: new emphasis on the role of macroprudential regulations more targeted than MP to achieve financial stability MP may at times face tradeoffs: e.g. low inflation and high HH debt Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

4 In this paper We build a medium scale, small-open-economy DSGE model with real-financial linkages: balance sheet positions of banks, HHs, and firms affect funding/lending conditions and the real economy macroprudential policies: capital requirements; LTV nominal and real frictions: monetary policy and propagation We use the model to analyze transmission of financial shocks (e.g. exuberance, risk premium) effects of macroprudential policies on real and financial variables Main finding: LTV most effective and least costly in reducing HH debt, followed by capital regulations and MP. Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

5 Literature: Balance sheet and bank capital channels Firm and HH leverage key determinant of borrowing conditions Carlstrom and Fuerst (1997), Bernanke, Gertler and Gilchrist (1999), Kiyotaki and Moore (1997), Iacoviello (2005) amplification due to "financial accelerator" Firms/HHs Banks Assets Liabilities Assets Liabilities Capital Bank loans Loans Funding Housing Net worth Securities Equity Capital position of banks key for funding and lending conditions Holmstom and Tirole (1997), Meh and Moran (2010), Gertler and Karadi (2010), Davis (2010), Gerali et al. (2010) further amplification due to "adverse feedback loop" Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

6 Model Small-open-economy DSGE model with financial frictions HH lending to banks, and bank lending to HHs/firms involve monitoring costs (Curdia and Woodford, 2011) spreads in funding/lending rates depend on bank and borrower leverage (Bernanke et al., 1999; Davis, 2010) macroprudential policies feed into spreads financial shocks (e.g. capital quality; spreads; exuberance) Other key features nominal frictions: price/wage stickiness; indexation real frictions: habit formation; utilization and investment adj. costs open economy: extended UIP condition; partial pass-through monetary policy: Taylor rule Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

7 Brief overview of the model Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

8 Banks Use deposits and bank capital to fund lending to impatient households and entrepreneurs P I,t b I,t + P E,t b E,t = D t + A t Pay dividends to patient HHs; dividend smoothing based on adj. costs (Jermann and Quadrini, 2012) Bank loans are modeled as perpetuities with exponentially decaying coupon payments (Woodford, 2001) bank loan issued in t 1 priced in period t as (δ/π t ) P I,t ; allows recursive formulation for banks cash-flow (ABS) Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

9 Monitoring costs and lending rates Banks incur monitoring costs based on market value of bank assets/lending (Curdia and Woodford, 2011) captures "bad loans"/default; cost of purchasing default insurance costs increase with borrower leverage spread similar to BGG Monitoring costs for bank lending to HHs and bank funding are modeled similarly modeling choice captures main themes in literature while avoiding technical issues (long-term borrowing; risk-averse borrowers) Bank borrow short-term, lend long-term (modeled as perpetuities with exponentially decaying coupon payments as in Woodford, 2001) long-term lending rates depend on: current and future policy rates and bank/borrower leverage R I,t = ( 1 δ ) I ( ) s δi E t [ R t+s + Ψ d,t+s + Ψ I,t+1] R I s=0 R I Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

10 Macroprudential policy and spreads Regulatory LTV policy feeds into monitoring costs on household loans ( ) (1 mt ) q Ψ I,t = h,t h I,t f n I,t monitoring costs increase as equity falls short of regulated equity translates quantity restrictions into spreads Similarly, bank capital regulations impact funding spread ( ) γt [ω Ψ d,t = I P I,t b I,t + ω E P E,t b E,t ] f regulations do not necessarily bind in the short-run A t Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

11 Calibration Main targets for steady-state: 3% risk-free rate; 14 bps. funding spread 240 and 200 bps. overall spread on HH and business loans mortgage debt / impatient HH housing = 95% (b I /h I ) mortgage debt / total housing = 30% (b I /h) business debt / entrep. capital = 0.5 (b E /k E ) business debt / total capital = 0.25 (b E /k) bank capital ratio = 10% Parameters determining dynamics: nominal and real frictions: ToTEM financial frictions: HRAM; MAG report (BIS) Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

12 Tightening in regulatory LTV: transmission mechanism Increase in current and future monitoring costs of banks, Ψ I,t+k lending rate to HHs, R I impatient HH borrowing, b I ; demand for housing and consumption house price, q h ; residential investment Spillover to non-residential sector and amplification demand for non-res. investment, capital price, q k bank capital = bank funding/lending rates Moderating factors π policy rate, R patient HH increase demand: c P, h P, k P currency depreciation exports Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

13 Tightening in regulatory LTV Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

14 Transmission mechanism for capital req. and MP Increase in bank capital requirements direct impact on borrowing cost of entrepreneurs as well as HHs larger decline in business loans and non-res. investment "broad" tool to reduce HH debt Increase in policy rates adverse impact on patient HH consumption demand and exports "blunt tool" to reduce HH debt Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

15 Comparing effects of different policies Comparing peak effects of macroprudential and monetary policies (in pp) 5 pp in LTV 1 pp in cap. req. 100 bp in policy rate Output Consumption Business inv Residential inv Inflation House price Household debt LTV most effective in reducing HH debt with least output cost; relative to capital req. and monetary policy. Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

16 Optimal LTV policy and policy coordination Scenario: low π ("external headwinds") and high HH debt ("exuberance") Optimal regulatory LTV ratio: slightly higher than 0.95 what s the financial stability concern? pecuniary externality: spreads depend on aggregate variables "exuberance shocks": affect expected return on assets Instead consider high impact event (crisis) due to high HH debt HH debt level after 5 years determines shock intensity optimal policy: MP loosening and LTV tightening Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

17 Alternative modeling of long-term mortgages Debt accumulation: D t = (1 κ) D t 1 + L t Average borrowing rate on outstanding mortgages, Rt M : D t Rt M = (1 Φ) (D t L t ) Rt 1 M + [L t + Φ (D t L t )] Rt F, where Φ : share of existing loans refinanced at current fixed rate, Rt F New borrowing: L t = φp t q h,t i hi,t + γ [P t q h,t (1 δ h ) h I,t 1 (1 κ) D t 1 ] + ε l,t If κ = 1 and γ = φ, then constraint becomes similar to Iacoviello (2005) D t = φp t q h,t h I,t Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

18 Smaller impact of LTV on HH debt-to-gdp ratio Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19

19 Conclusion Small open-economy DSGE model with nominal, real and financial frictions captures main themes in RFL literature while keeping model tractable can be used to analyze effects of macroprudential policy Related work/possible extensions: flow vs. stock of HH debt small probability/high impact event default and risk-taking channel of MP term-premium vs. credit risk (QE) Some final thoughts on policy implications: MP tightening may increase HH debt-to-gdp ratio (in SR) significant tightening in LTV required to reduce HH debt Alpanda, Cateau, Meh (Bank of Canada) ()Macroprudential - Monetary Policy Model October / 19