Macroprudential policies for a small open economy Aino Silvo Fabio Verona Bank of Finland, Monetary Policy and Research Department aino.silvo@bof.fi fabio.verona@bof.fi 3 rd Research Conference of the CEPR Network on Macroeconomic Modelling and Model Comparison Frankfurt am Main, 13-14 June 2019 The views expressed are ours and do not necessarily reflect the views of the Bank of Finland
The need of macro models ... ... for macroprudential policy simulations The global financial crisis added financial frictions to the agenda of macroeconomic research At the same time, new tasks such as macroprudential supervision, required central banks to employ structural models that facilitate the evaluation of macroprudential policies These requirements set in motion the development of the Aino 3.0 model, which builds on its predecessor – the Aino 2.0 model 1 From the point of view of Finland, monetary policy is exogenous and macroprudential is an active domestic policy 1 Kilponen, Ripatti, Orjasniemi and Verona (2016), “ The Aino 2.0 model ”, Bank of Finland Research Discussion Paper 16/2016
Macroprudential policy simulations Do macroprudential policies help making the economy more stable ( i.e. , less volatile)? Are borrower-based instruments (LTV ratio) more effective in stabilizing the economy than capital-based tools (risk weights)? They stabilize the “financial” cycle at the expenses of larger real business cycle fluctuations
Features of the Finnish housing market The housing sector ◮ Residential construction: an important driver of the Finnish business cycle ◮ 2/3 of aggregate household wealth is in residential real estate Household indebtedness rate: 128% ◮ Considered as one of the key macroeconomic vulnerabilities in the Finnish economy Transmission of monetary policy ◮ Direct interest rate channel through variable lending rates ◮ Indirect balance sheet channel through collateral values and lending spreads Homeownership rate: 67%
The Aino 3.0 model
The Aino 3.0 model Households Iacoviello (AER 2005) Patient / savers, share ω h = 0 . 66 , discount at rate β P ◮ Consume, work, hold houses, save in bonds and bank deposit Impatient / borrowers, share 1 − ω h , discount at rate β I < β P ◮ Consume, work, hold houses, borrow from banks to finance the housing purchase ◮ (Always binding) collateral constrained: the loan amount ( BL H , new ) is a t + 1 fraction ( θ H t ) of the market value of the new housing being purchased � � 1 − δ H � � BL H , new ≤ θ H t P H H t + 1 − H t t t + 1
The Aino 3.0 model Multiperiod mortgage loans Kydland, Rupert and Sustek (IER 2016) and Garriga, Kydland and Sustek (RFS 2017) t + BL H , new Stock of outstanding debt: BL H t + 1 = ( 1 − γ t ) BL H where t + 1 γ t ≤ 1 is the effective amortisation rate � γ t +( 1 − τ r t ) r H � BL H t where r H Period mortgage payment: MP t = t − 1 t is the (variable) interest rate on home loans and τ r t is the tax deduction on the interest rate � � BL H , new BL H , new ( γ t ) α + Effective amortisation rate: γ t + 1 = 1 − t + 1 t + 1 t + 1 κ BL H BL H t + 1 where 0 < κ ≤ 1 is the initial amortisation rate of new loans and 0 ≤ α ≤ 1 governs the evolution of the effective amortisation rate ◮ α = 0 and κ = 1 yields the one-period loan framework ◮ α = 1 gives the constant amortisation (decaying coupon) framework with an amortisation rate γ t = κ
The Aino 3.0 model Banks Gerali, Neri, Sessa and Signoretti (JMCB 2010) (Exogenous) bank capital requirement: the bank has to pay a cost whenever the capital-to-risk-weighted-assets ratio BL NFC + Φ H BL H � K b / � is different from the target value v b t t = v b = 8 %(+ 2 . 5 %) ◮ Mandatory capital requirement: v b t = v b + χ v � � B t Y t − B ◮ Countercyclical capital buffer: v b Y ◮ Risk-weight requirement associated with mortgage loans: Φ H
The Aino 3.0 model Banks Monopolistic competition and sticky loan rates: different interest rate pass-through for NFC and housing mortgage loans t + 1 + c 31 ˜ r NFC r NFC r NFC R b ε NFC ˜ = c 11 ˜ t − 1 + c 21 E t ˜ t − c 41 ˜ t t r H r H r H t + 1 + c 32 ˜ R b ε H ˜ t = c 12 ˜ t − 1 + c 22 E t ˜ t − c 42 ˜ t � reg � ˜ t = ˜ ˜ t + 1 − ˜ R b k b v b R t − c 5 bl t + 1 − ˜ t reg NFC � H � ˜ t + 1 = c 6 ˜ φ H t + ˜ t + 1 + c 7 bl bl bl t + 1 Note: ˜ R t is the “ECB” interest rate, which is exogenous
Steady state Data † Variable Model C / Y 0.76 0.77 I / Y 0.27 0.27 X / Y 0.54 0.54 MC / Y 0.17 0.18 MI / Y 0.40 0.40 MX / Y 0.51 0.51 housing investment / Y 0.09 0.10 non-financial corporations loans-to-gdp ratio 1.47 1.47 mortgage loans-to-gdp ratio 1.77 1.40 spread (corporate loan rate - risk-free rate), annual (%) 1.55 1.55 spread (mortgage loan rate - risk-free rate), annual (%) 1.62 1.62 † Sample period: 1996Q1 – 2018Q3
Macroprudential policy simulations Do macroprudential policies help making the economy more stable ( i.e. , less volatile)? Are borrower-based instruments (LTV ratio) more effective in stabilizing the economy than capital-based tools (risk weights)? Two simulation scenarios ◮ House demand shock which generates a boom in house prices ◮ Anticipated and permanent low for long Euro Area monetary policy
Macroprudential policy simulations House demand shock (ar = 0.75) LTV=0.9 & RW=0.15; LTV=0.9, RW=0.5; LTV=0.8, RW=0.15
Macroprudential policy simulations House demand shock (ar = 0.75) LTV=0.9 & RW=0.15; LTV=0.9, RW=0.5; LTV=0.8, RW=0.15
Macroprudential policy simulations House demand shock (ar = 0.75) LTV=0.9 & RW=0.15; LTV=0.9, RW=0.5; LTV=0.8, RW=0.15
Macroprudential policy simulations House demand shock (ar = 0.75) Percentage difference with respect to the baseline calibration (LTV=0.90 & RW=0.15) σ gdp σ h . inv . σ h . prices σ ml / gdp LTV=0.90 & RW=0.50 0.5 0.4 0.3 5.7 LTV=0.80 & RW=0.15 51.7 5.8 1.9 -14.9
Macroprudential policy simulations Anticipated and permanent low for long Euro Area monetary policy LTV=0.9 & RW=0.15; LTV=0.9, RW=0.5; LTV=0.8, RW=0.15
Macroprudential policy simulations Anticipated and permanent low for long Euro Area monetary policy LTV=0.9 & RW=0.15; LTV=0.9, RW=0.5; LTV=0.8, RW=0.15
Macroprudential policy simulations Anticipated and permanent low for long Euro Area monetary policy LTV=0.9 & RW=0.15; LTV=0.9, RW=0.5; LTV=0.8, RW=0.15
What’s next Better calibration of the elasticities / parameters that drive the dynamics How to quantify the gains / losses Which variables does/should the macroprudential authority care about? Which shocks are important (to respond to)? SVAR / DSGE model for foreign variables to improve e.g. the monetary policy transmission mechanism Effects of other macroprudential policy tools Estimate the model . . .
Extra slides
Aino model vintages @ the Bank of Finland 1 st version – Aino – August 2004 ◮ Dynamic general equilibrium model à la Gertler (1999) ◮ Calibrated, focus on fiscal and demographic issues 2 nd version – eAino – 2010-2012 ◮ Open economy DSGE model similar to e.g. the Riksbank Ramses model or the ECB New Area-Wide model ◮ Estimated, focus on business cycle analysis in a small open economy ◮ Used for forecasting the Finnish economy 3 rd version – Aino 2.0 – 2014-2016 ◮ eAino augmented with a banking sector and corporate loans ◮ Estimated, used for forecasting the Finnish economy 4 th version – Aino 3.0 – 2017-... ◮ Aino 2.0 augmented with residential housing construction and long-term mortgage loans ◮ Calibrated, focus on macroprudential analysis
The Aino model evolution eAino
The Aino model evolution Aino 2.0
The Aino model evolution Aino 3.0
The Aino model evolution Aino 3.0
Model dynamics IRFs to a monetary policy shock (ar = 0.95) black lines: LTV=0.9 & RW=0.15; red lines: LTV=0.9, RW=0.5; blue lines: LTV=0.8, RW=0.15
Model dynamics IRFs to a house demand shock (ar = 0.75) black lines: LTV=0.9 & RW=0.15; red lines: LTV=0.9, RW=0.5; blue lines: LTV=0.8, RW=0.15
Model dynamics IRFs to a capital productivity shock (ar = 0.14) black lines: LTV=0.9 & RW=0.15; red lines: LTV=0.9, RW=0.5; blue lines: LTV=0.8, RW=0.15
Model dynamics IRFs to a foreign demand shock (ar = 0.95) black lines: LTV=0.9 & RW=0.15; red lines: LTV=0.9, RW=0.5; blue lines: LTV=0.8, RW=0.15
Model dynamics IRFs to a government spending shock (ar = 0.82) black lines: LTV=0.9 & RW=0.15; red lines: LTV=0.9, RW=0.5; blue lines: LTV=0.8, RW=0.15
Model dynamics IRFs to a LTV ratio shock (ar = 0.90) black lines: LTV=0.9 & RW=0.15; red lines: LTV=0.9, RW=0.5; blue lines: LTV=0.8, RW=0.15
Model dynamics IRFs to a price markup shock (ar = 0) black lines: LTV=0.9 & RW=0.15; red lines: LTV=0.9, RW=0.5; blue lines: LTV=0.8, RW=0.15
Recommend
More recommend
