Fiscal Multipliers and Financial Crises

Fiscal Multipliers and Financial Crises Miguel Faria-e-Castro New York University June 20, 2017 1 st Research Conference of the CEPR Network on Macroeconomic Modelling and Model Comparison 0 / 12

Fiscal Multipliers and Financial Crises Fiscal policy response to the 2008 financial crisis and Great Recession 1. Conventional stimulus (purchases and transfers) 2. Financial sector interventions (equity injections and guarantees) Large debate on the effectiveness and composition of the response 1 / 12

Approach and Results 1. Structural model of fiscal policy Potential stabilization roles for each of the tools State dependent effects of shocks and policies 2 / 12

Approach and Results 1. Structural model of fiscal policy Potential stabilization roles for each of the tools State dependent effects of shocks and policies 2. Quantitative Exercise Combine calibrated model with data on fiscal response Estimate structural shocks given fiscal policy response Study counterfactuals Crisis and Great Recession without fiscal response 3. Results: Aggregate consumption falls by 50% more without policy response Transfers and equity injections most important Fiscal multipliers extremely state dependent New transmission channels for fiscal policy 2 / 12

Model Nominal Rigidities = Government purchases Incomplete Markets = Transfers (Frictional) Financial Sector = Bank Recaps. Credit Risk & Default = Credit Guarantees 3 / 12

Model Nominal Rigidities = Government purchases Incomplete Markets = Transfers (Frictional) Financial Sector = Bank Recaps. Credit Risk & Default = Credit Guarantees Government Transfers, T b Guarantees, s d Recaps., s k Purchases, G Savers Risky Deposits Banks Leverage Constraint Risky Loans Borrowers LTV Constraint Consumption Labor Labor Consumption Firms/Goods Market Housing 3 / 12

Impulse and Propagation Aggregate shocks: 1. TFP A t 2. Financial shock σ t Household Default Rate t = f ( LTV t, σ + t ) + 4 / 12

Impulse and Propagation Aggregate shocks: 1. TFP A t 2. Financial shock σ t Household Default Rate t = f ( LTV t, σ + t ) + Financial shock: defaults 4 / 12

Impulse and Propagation Aggregate shocks: 1. TFP A t 2. Financial shock σ t Household Default Rate t = f ( LTV t, σ + t ) + Financial shock: defaults 1. Bank equity 4 / 12

Impulse and Propagation Aggregate shocks: 1. TFP A t 2. Financial shock σ t Household Default Rate t = f ( LTV t, σ + t ) + Financial shock: defaults 1. Bank equity 2. If bank constraint binds spreads rise, lending falls 3. Disposable income for borrowers 4. If borrower constraint binds aggregate consumption 4 / 12

State Dependence: Financial Shock with Low Leverage GDP Cons. Borrower % change from t-4 0.5 0-0.5-1 -4-2 0 2 4 6 % change from t-4 0-2 -4-4 -2 0 2 4 6 % change from t-4 5 0-5 -10 Value of the Bank change from t-4 1.5 1 0.5 Bank Cost of Funds -4-2 0 2 4 6 Quarters after crisis -4-2 0 2 4 6 Quarters after crisis 5 / 12

State Dependence: Financial Shock with High Leverage GDP Cons. Borrower % change from t-4 0-1 -2-4 -2 0 2 4 6 % change from t-4 0-2 -4-6 -4-2 0 2 4 6 % change from t-4 0-10 -20 Value of the Bank change from t-4 2 1.5 1 0.5 Bank Cost of Funds Low lev. Hi lev. -4-2 0 2 4 6 Quarters after crisis -4-2 0 2 4 6 Quarters after crisis 6 / 12

Quantitative Exercise 1. Calibrate model to U.S. pre-crisis Match moments related to household and bank balance sheets Calibration 7 / 12

Quantitative Exercise 1. Calibrate model to U.S. pre-crisis Match moments related to household and bank balance sheets Calibration 2. Data on fiscal policy response Ω T = { } G t, Tt b, st k, st d T t=0 Collected from a variety of sources Data 3. Estimate {A t, σ t } T =2015Q4 t=2000q1 by making model match data given ΩT What are the sequences {Ât, ˆσ t } T t that make model match data T = {C t, spread t } T t given observed fiscal policy response 7 / 12

Main Counterfactual: No Fiscal Policy 2 Consumption 250 Lending Spread % dev. from trend 1 0-1 -2 200 150 100 50 Counterfactual Data -3 0 2007Q1 2008Q3 2013Q4 2007Q1 2008Q3 2013Q4 8 / 12

Policy Decomposition 2 1 Aggregate Consumption Data No Purchases No Transfers No Bank Recaps % deviation from trend/ss 0-1 -2-3 -4 2007Q1 2008Q3 2012Q4 9 / 12

Time Series for Fiscal Multipliers 10 / 12

State Dependent Multipliers: Mechanism Two channels: 11 / 12

State Dependent Multipliers: Mechanism Two channels: 1. Borrower constraint conventional MPC channel 11 / 12

State Dependent Multipliers: Mechanism Two channels: 1. Borrower constraint conventional MPC channel 2. Borrower + Bank constraints new channel 11 / 12

State Dependent Multipliers: Mechanism Two channels: 1. Borrower constraint conventional MPC channel 2. Borrower + Bank constraints new channel Transfers house prices (only when borrowers are constrained) Default rates fall, banks post fewer losses Lending, spreads (only when banks are constrained) 11 / 12

Conclusion This Paper Analysis of fiscal policy response to the Great Recession Structural Model + Data 12 / 12

Conclusion This Paper Analysis of fiscal policy response to the Great Recession Structural Model + Data Contribution Conventional stimulus and financial sector interventions Important for normative analysis Quantitative evaluation New transmission channels for fiscal policy Household-bank balance sheet interactions State dependent effects 12 / 12

Appendix 1 / 11

Borrowers: Debt and Default Face value Bt 1 b, Fraction γ matures every period Family construct (Landvoigt, 2015) 1. Borrower enters period with states h t 1, B b t 1 2. Continuum of members i [0, 1], each with h t 1, B b t 1, ν t (i) where ν t (i) F b t [0, ) 3. Each member i can: 3.1 Repay maturing debt γb b t 1, and keep houses worth ν t(i)p th t 1 or 3.2 Default on maturing debt, lose collateral Back 2 / 11

Borrower Family Problem Vt b (Bb t 1, h t 1) = max ct b,nb t,ht,bb t,ι(ν) subject to budget constraint { log(c t) v(n t) + ξ b log(h t) + β ct b + γ Bb t 1 [1 ι(ν)]dft b (ν) + p t h t Π } t }{{}{{} house purchase debt repayment (1 τ)w t nt b + Qt b Bt b,new }{{} new debt and borrowing constraint + p t h t 1 ν[1 γι(ν)]dft b (ν) T t + } {{ } sale of non-forecl. houses B b,new t mp t h t [ } 1 E t(vt+1 b )1 αb] 1 α b T b t }{{} Transfers Back 3 / 11

Borrower Default Default iff ν ν t, ν t = Bb t 1 Π t p t h t 1 Loan-to-Value F b t = Beta(1, σ b t ) σt b two-state Markov Mean preserving spread 5 4.5 4 3.5 3 Normal Crisis f b t, pdf 2.5 2 ν 1.5 1 0.5 0 0.7 0.75 0.8 0.85 0.9 0.95 1 ν Lenders earn (per unit of debt) Resource Cost {}}{ ν t Zt loans = (1 γ)qt b +γ 1 Ft b (νt ) + (1 λ b ) ν ptht 1 df b }{{}}{{} 0 Bt 1 not matured b /Πt t repaid }{{} foreclosed 4 / 11

Financial Intermediaries Fixed income portfolios, maturity transformation, risky deposits Fraction 1 θ of earnings paid out as dividends every period Invest in loan securities b t, raise deposits d t Problem for intermediary j [0, 1] with current earnings e j,t Vt k (e j,t ) = max }{{} b j,t,d j,t current mkt value subject to [ Λ s t,t+1 (1 θ)e j,t + E t }{{} Π t+1 dividend flow of funds : Q b t b j,t = θe j,t (1 + s k t ) + Q d t d j,t max { 0, V k t+1(e j,t+1 ) }] } {{ } ex-dividend value [ Λ s capital req. : κqt b t,t+1 b j,t E t max { 0, V Π t+1(e k j,t+1 ) }] t+1 LoM earnings : e j,t+1 = u j,t+1 Zt+1 loans b j,t d j,t Back 5 / 11

Financial Intermediaries u j,t F d [u, ū] Default iff [0,1] u j,t < u t d j,t 1 Zt loans b j,t 1 Leverage Aggregation representative bank [ Λ s t,t+1 E t max { 0, V Π t+1(e k j,t+1 ) }] dj Φ t θe t t+1 Spreads reflect Credit Risk + Current + Future binding constraints Long-term debt Pecuniary Externalities Financial Accelerator Payoff per unit of deposits, Z deposits t = st d +(1 st d ) }{{} guaranteed 1 F d (ut ) }{{} repaid u t + (1 λ d ) u Z t loans Bt 1 b df d D 0 t 1 }{{} liquidated Back 6 / 11

Closing the Model Standard DSGE model w/ nominal rigidities Producers Phillips Curve producers Savers Euler Equation (IS) savers Housing in fixed supply, Central Bank Taylor Rule h t = 1 1 Q t = 1 Q Aggregate resource constraint, [ Πt Π ] φπ [ ] φy Yt Y C t+g t+dwl Bank Default t+dwl Household Default t = A }{{} tn t [1 d(π t)] }{{} =Y t Menu Costs Back 7 / 11

Fiscal Authority Budget constraint, τy t + T t + Q t Bt g Ḡ Bg t 1 = Net Cost from Discretionary Measures Π t }{{ t } Standard Surplus Fiscal rule for taxes, ( B g ) t 1 T t = φ τ log B g Net Cost from Discretionary Measures, (G t Ḡ) + χt b t + s k t θe t + s d t D t 1 Π t (1 Recovery Rate t ) Back 8 / 11

Calibration 1. Crises σt b = [σt b,normal, σt b,crisis ] T and P σ = [.995 ].005.2.8 2. Households Target Target Parameter Fraction Borrowers Parker et al. (2013) χ = 0.45 Avg. Maturity 20 quarters γ = 0.05 Max LTV Ratio 80% m = 0.203 Debt/GDP 80% ξ = 0.0347 Avg. Delinquency Rate 2% σ b,normal = 12.93 3. Banks F d (u) = uσ u σ ū σ u σ Target Target Parameter Book Leverage 10 κ = 0.1 Payout Rate 20% θ = 0.8 Avg. Lending Spread 2% ϖ = 0.0105 CDS-Implied Prob. Def. 2% in recessions u = 0.91, σ d = 1 Back 9 / 11

Fiscal Policy Response to the Great Recession % of 2007 GDP 0 2 4 6 8 2007q1 2008q3 2013q4 Govt. Purchases Equity Injections Transfers Guarantees Back 10 / 11

Smoothed Shocks 11 / 11