A Tale of Two Stimulus Payments: 2001 vs 2008

Transcription

1 A Tale of Two Stimulus Payments: 2001 vs 2008 Greg Kaplan Princeton University & NBER Gianluca Violante New York University, CEPR & NBER American Economic Associa-on Annual Mee-ng January 5, 2013

2 Fiscal stimulus payments Small, anticipated, temporary, (almost) lump-sum payments Frequently used instrument to stimulate spending during recessions Two recent episodes: 2001 (EGTRRA): $38b total payout, 1.7% quarterly GDP 2008 (ESA): $79b total payout, 2.2% quarterly GDP Two objectives: 1. Alleviate household economic hardship 2. First-round impulse to fiscal multiplier

3 Empirical evidence Exploit randomization in timing of receipt stimulus payments to estimate consumption responses 2001 (EGTRRA): Johnson et al. (2006, JPS) 2008 (ESA): Parker et al. (2013, PJSM) Findings for non-durable consumption: Strong response: around 25% spent in quarter of receipt Smaller point estimates in 2008: around 5pp lower

4 Challenge for standard theory Permanent Income Hypothesis (PIH) Zero response to anticipated transitory windfall

5 Challenge for standard theory Permanent Income Hypothesis (PIH) Zero response to anticipated transitory windfall Standard Incomplete Markets Model (SIM) Consumption response from liquidity constrained households Disciplined by data on net worth <10% households constrained è rebate coefficient < 5%

6 Challenge for standard theory Permanent Income Hypothesis (PIH) Zero response to anticipated transitory windfall Standard Incomplete Markets Model (SIM) Consumption response from liquidity constrained households Disciplined by data on net worth <10% households constrained è rebate coefficient < 5% Kaplan and Violante (2013) (KV13) 2 asset model with transaction costs Generates wealthy hand-to-mouth households Disciplined by data on liquid and illiquid assets Parameterized to match 2001 SCF data è rebate coefficients closer to empirical estimates of JPS

7 Our objective Use KV13 framework to shed light on the difference between the consumption responses to the fiscal stimulus policies of 2001 and 2008, as measured by JPS and PJSM

8 2001 vs 2008: 5 key differences Design of policy D1: Payment size $500 $1000 D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession

9 2001 vs 2008: 5 key differences Design of policy D1: Payment size $500 $1000 D2: Phasing out none starting at $75K D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession

10 2001 vs 2008: 5 key differences Design of policy D1: Payment size $500 $1000 D2: Phasing out none starting at $75K D3: Minimum income none at least $3K Economic Environment E1: Tax reform E2: Size of recession

11 2001 vs 2008: 5 key differences Design of policy D1: Payment size $500 $1000 D2: Phasing out none starting at $75K D3: Minimum income none at least $3K Economic Environment E1: Tax reform Bush tax cuts none E2: Size of recession

12 2001 vs 2008: 5 key differences Design of policy D1: Payment size $500 $1000 D2: Phasing out none starting at $75K D3: Minimum income none at least $3K Economic Environment E1: Tax reform Bush tax cuts none E2: Size of recession short and shallow long and deep

13 Model Baumol-Tobin model of money demand integrated into a partial equilibrium, life-cycle, incomplete markets economy Finite horizon Idiosyncratic fluctuating labor income while working Social security benefits when retired Epstein-Zin-Weil preferences defined over Non-durable consumption Housing services Two assets: Liquid asset with low return (r m ) Illiquid with higher return (r a ) and housing service flow (³) Depositing or withdrawing from illiquid asset requires paying a fixed transaction cost ( ) Unsecured borrowing in liquid asset at rate r b > r m

14 Parameterization Preference and risk parameters as in KV13 (2001 steady-state): CRRA = 4 IES = 1.5 Discount factor ( ) to match median illiquid wealth Earnings risk is unit-root

15 Parameterization Real asset returns (after tax, annual) as in KV13: Liquid assets: cash; money market, checking, savings and call accounts; directly held mutual funds, stocks, bonds, T-bills; net of revolving debt on credit card balances: 1.48% Illiquid assets: net housing wealth, retirement accounts, life insurance, CDs and savings bonds: 2.29% Housing service flow: maintenance, insurance, property taxes, interest; imputed rent, tax deductability: 4% of stock

16 Parameterization Transaction cost ( ) and borrowing rate (r b ) Match fraction of wealthy and poor hand-to-mouth (2001 SCF) In KV13 outline strategy for identifying a lower bound: 20% - 40% of US households are hand-to-mouth 1/3 of these are poor, 2/3 are wealthy Target upper end of range: and r b =15.5% In KV13 we target middle of range: and r b =10%

17 Experiment Replicate 2001 tax rebate episode in model Economy is in steady-state when hit with three pieces of news: 1. Recession of depth and length of 2001 downturn 2. Tax reform that mimics EGTRRA 3. $500 tax rebate to half the population immediately, and to half the population in the following quarter Compute transition and run JPS/PJSM regressions on modelgenerated panel data to compute rebate coefficients Repeat experiment with 5 differences for 2008

18 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession 2008: D1+D2+D3 + E1+E2

19 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession 2008: D1+D2+D3 + E1+E2

20 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession 2008: D1+D2+D3 + E1+E2

21 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession 2008: D1+D2+D3 + E1+E2

22 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession 2008: D1+D2+D3 + E1+E2

23 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession : D1+D2+D3 + E1+E2

24 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession : D1+D2+D3 + E1+E

25 Model rebate coefficients: 2001 vs 2008 Baseline calibration r m = 15.5% KV13 calibration r m = 10% Design of policy D1: Payment size D2: Phasing out D3: Minimum income Economic Environment E1: Tax reform E2: Size of recession : D1+D2+D3 + E1+E

26 Conclusions Use KV13 framework to shed light on differences in consumption responses to the fiscal stimulus policies of 2001 and differences in policy design 2 difference in economic environment Combined effect: consumption response lower in 2008 by 1/3 Most important effect is the larger average payments in 2008