On the Merits of Conventional vs Unconventional Fiscal Policy Matthieu Lemoine and Jesper Lindé Banque de France and Sveriges Riksbank The views expressed in this paper do not necessarily reflect those of Banque de France and Sveriges Riksbank. ESCB Research Cluster, /8 Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Main messages Relative merits of conventional vs unconventional fiscal policy In the context of high debt levels and monetary constraints (ZLB), recent litterature has promoted unconventional fiscal policy (UFP) One important proposal is a gradual increase of the sales tax. See Correia et al. (3, AER) for a closed economy analysis and Fahri et al. (4, ReStud) for open economy extension (fiscal devaluation). We compare such a policy with conventional fiscal policy (CFP) based on hikes of government investment Data suggest historically low levels of public and private infrastructure investment in recent years in many EA countries (IMF issued support for increased public investment in Germany, May 4th). Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Main findings Conventional fiscal policy is more robust to adding realistic frictions UFP is appealing in a stylized model with sticky prices and flexible wages. Both policies are expansionary, but UFP has the virtue of reducing public debt. Advantage of UFP does not necessary hold up in a richer TANK model with sticky wages Contrary to conventional policy, UFP is not expansionary anymore unless labor (or capital) income taxes are adjusted aggressively. Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 3 / 3
Presentation outline Stylized Model Impulses to Gradual Sales Tax Hike Impulses to Higher Government Investment Analysis in a Fully-Fledged Model Tentative Conclusions Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 4 / 3
Stylized Model Overview A standard log-linearized version of the New Keynesian DSGE model with a lower bound constraint on interest rates following Eggertsson and Woodford (3) Separability between consumption and labor Production function with fixed private capital and with public capital subject to time-to-build Sticky prices, flexible nominal wages Government consumes/invests part of final domestic good Sales taxes exogenous Labor income tax rule stabilizes gov t debt Standard calibration intended to be relevant for US and EA Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 5 / 3
Stylized Model Setting of simulations of sales tax hikes Setting of the two experiments: announced increase of sales taxes by pp of GDP, with aggressive or non agg. response of labor tax, within -quarter liquidity trap (achieved with taste shock ν t ).5 Sales tax.5 Quarter Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 6 / 3
Stylized Model Key ingredients driving responses to sales tax hikes Key equation: the dynamic IS curve extended with sales tax τ C,t (C t C ν t ) /σ + τ C,t = βe t + i t + π t+ (C t+ C ν t+ ) /σ + τ C,t+ Key responses As inflation is sticky, slow response of real interest rate Because of expected increase of τ C,t, big incentive to substitute current consumption to future consumption Inflation stimulated by the increased output gap (Phillips curve) Extra receipts from tax hike and its expansionary effect imply fall of debt or of labor tax depending on fiscal rule Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 7 / 3
Impulse Responses to Gradual Sales Tax Hike percent of baseline GDP hike in stylized model Figure. Impulses to Sales Taxes in Normal Times and in a Quarter Liquidity Trap Non-agressive Tax Rule Agressive Tax Rule.5.5 Normal Times Potential Liquidity Trap.5.5 -.5 -.5 Real Interest Rate (APR) Real Interest Rate (APR) - Labor Income Tax - - -3 Govt. Debt (share of trend GDP) - -4 - Labor Income Tax - - -3 Govt. Debt (share of trend GDP) - -4-6 -6 - Quarter - Quarter Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 8 / 3
Stylized Model Setting of simulations of gov t investment stimulus Setting of the two experiments: increase of gov t investment by pp of baseline GDP at the onset of a -quarter liquidity trap. Public Infrast. Investment (GDP share).8.6.4. Quarter Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 9 / 3
Stylized Model Key ingredients driving responses to gov t investment stimulus Key equations: supply/demand role of gov t investment G I,t [ Y t = Z t K ν P KG,t ν ] α N α t = C t + G C,t + G I,t Key responses Fall of real interest rate because of rising demand and inflation expectations Potential real rate only start to rise in the phasing out period (with flex. price, crowding in of consumption at that time) gap boosted by this fall of the real interest rate gap Here, demand channel reinforced by impact of gov t investment on permanent income (smaller role for impact on marginal costs) Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Impulse Responses to Higher Government Investment percent of baseline GDP hike in stylized model Figure 3. Impulses to Public Invest. in Normal Times and in a Quarter Liqu. Trap Non-agressive Tax Rule Agressive Tax Rule Normal Times Liquidity Trap Potential - Real Interest Rate (APR) - Real Interest Rate (APR) - Labor Income Tax - Labor Income Tax - Govt. Debt (share of trend GDP) 5 - Govt. Debt (share of trend GDP) 5-5 Effective Capital Stock -5 Effective Capital Stock.5.5 Quarter Quarter Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Analysis in the Fully-Fledged Model Overview of model DSGE model of Erceg and Lindé (3), building on CEE/SW Endogenous private capital Nominal and real rigidities CEE (5), SW (3, 7): Staggered price and wage contracts, dynamic indexation External habit persistence in consumption CEE type of investment adjustment costs Hand-to-mouth households following EGG (6) Financial accelerator mechanism; CMR (7) variant of BGG (999) Other aspects as in stylized model, but more realistic modelling of tax bases and positive steady state debt Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Impulse Responses to Both Shocks percent of baseline GDP hike in fully-fledge model Sales tax hike: not anymore expansionary, unless aggressive tax rule Gov t investment stimulus: expansionary effects, as in the stylized case Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 3 / 3
Impulse Responses to Gradual Sales Tax Hike What drives the difference w.r.t. the stylized model? Sticky Price Model.5 Potential.5 -.5 Normal Times Liquidity Trap - - Sticky Price - Sticky Wage Model.5.5 -.5 - - Full Model.5.5 -.5 - - Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 4 / 3
Robustness checks Discounting of Euler equation: does not matter so much Expansionary effect of CFP robust to the productive degree of gov t spending and to time-to-build delay UFP more expansionary with an aggressive tax rule based on capital income tax GHH preferences: expansionary effect of UFP disappears even in stylized model (with non aggressive tax rule) Durables: could be in favor of UFP (work in progress) Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 5 / 3
Tentative Conclusions For an economy facing a deep recession and prolonged liquidity trap, there is a strong argument for temporarily increasing government spending on infrastructure spending. Such a policy would boost demand in the near-term which is useful, and potential output in the longer term when the economy is recovering. Given the slack of the Euro area, for its members with fiscal space this is still thus a strong case for fiscal stimulus. But for a country like the United States, which now experiences more normalized business cycle conditions, the macroeconomic argument for stimulus via infrastructure spending is much weaker. Benign effects of unconventional fiscal policy less certain, and dependent on grand tax bargains. The sales tax part of Abenomics not necessarily stimulative. Important lessons for ongoing empirical work (e.g. by D Acunto et al., 6). Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 6 / 3
Stylized Model Log-linearized representation IS curve (x t y t y pot t ) x t = x t+ t ˆσ(i t π t+ t r pot t ) Pricing schedule (NKPC) ( )) π t = βπ t+ t + κ mc (φ mc x t + τ N τ N,t τ pot N,t Potential output y pot t y pot t = φ mc ˆσ [g y g t + ( g y )ν c ν t Potential real interest rate r pot t r pot t ˆσ τ N τ pot N,t ˆσ + τ C τ C,t ] = ˆσ E t y pot t+ g y ˆσ E t g t+ g y νe t ν t+ + E t τ C,t+ ˆσ + τ c Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 7 / 3
Stylized Model Monetary policy specification Monetary policy rule i t = max { i, ( γ i ) (γ π π t + γ x x t ) + γ i i t } ZLB binding when log-linearized interest rate reach i The taste shock ν t follows a AR() process ν t = ( ρ ν )ν t + ε ν,t Add negative shock ε ν,t to make ZLB binding for quarters in the baseline scenario. Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 8 / 3
Stylized Model Fiscal Policy specification Gov t debt b G,t as a share of trend output (b G = ) evolves as ] b G,t = ( + r)b G,t + g y g t c y [τ C,t + τ C cy (y t g y g t ) s N [τ N,t + τ N (y t + φ mc x t )] τ t where (y t + φ mc x t ) equals real labor income, τ t lump-sum tax, and s N is the steady state labor share Fiscal policy rule based on the labor income tax τ N,t τ N = ϕ b b G,t + ϕ bb τ N,t where τ N,t is the labor income tax which keeps gov t debt b G,t fully stabilized. Non-agressive rule for a low value of ϕ b (ϕ bb = ) and complete stabilization rule for ϕ b = and ϕ bb = /s N Sales tax evolves according to a AR() process, written on error-correction form τ C,t = ρ τ, τ C,t ρ τ, τ C,t + ε C,t Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 9 / 3
Parameterization Calibration of key parameters Standard calibration intended to be relevant for the US and the euro area κ mc =., in line with empirical estimates for the U.S., e.g. GG (999) and Altig et al. () Assume standard simple rule for monetary policy unconstrained by ZLB (γ i =.7, γ π =.5, γ x =.5) Other parameters assume standard values; Frisch elasticity =.4, Labor share =.7, Government spending share =.3, log utility of consumption In the steady state, sales tax τ C =. as a compromise between levels in US and EA. Simplifying assumption that gov t debt b G = and τ t = τ =.6, so that income tax τ N =.33 when satisfying the steady state gov t budget constraint Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Stylized Model Sales tax hikes with and without discounting in Euler equation responses to sales tax hike in baseline stylized model Same in stylized model with discounted Euler equation Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Extending the stylized model with public investment Non-linear equations Instead of being fixed, total capital now affected by gov t capital: ( ) Y t = Z t K tot α t N α t Kt tot = (K P ) ϑ (K G,t ) ϑ with ϑ =.833 and α =.3, output elasticity of gov t capital stock equals.5 as in Leeper et al. (). Accumulation of gov t capital stock with a depreciation rate δ G =. K G,t = ( δ G )K G,t + I G,t Assumption of time-to-build, i.e. gov t spending turns into effective investment with delays (in a range of to 6 years) I G,t = 6 (G I,t 4 + G I,t 8 + G I,t + G I,t 6 + G I,t + G I,t 4 ) Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 / 3
Extending the stylized model with public investment Log-linearized implications Key equations of the log-linearized model remain unaltered, except the equation for yt pot which now becomes [ yt pot = g y ˆσ g t + ˆσ ( g ] y )ν c ν t τ N τ N,t ϕ mc +τ C τ C,t + +χ α (z t + α( ϑ)k G,t ) Total government spending (in log-linearized terms) equals g t = g C g Ct + g I g I,t where g Ct is government consumption, g C = G C /G and g I = g C. As gov t investment share of GDP equals 3 percent of GDP, against 3 percent for total gov t spending, we set g I =.3. Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 3 / 3
Impulse Responses to Higher Government Investment Robustness to alternative assumptions for non-aggressive rule Figure 3. Alternative Simulations of Impulses to Public Invest. 3 Benchmark calibration 3 Potential Normal Times Liquidity Trap 3 3 7 5 Public infrast. is not productive 5 9 9 3 7 3 3 3 7 5 Public infrast. is more productive 5 9 9 3 7 3 3 Public infrast. is sooner productive (- years) 3 3 Public infrast. is later productive (5- years) 3 Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 4 / 3
Analysis in the Fully-Fledged Model Parameterization We set the population share of the Keynesian households to optimizing households to.47, implies that the Keynesian households share of total consumption is about.3 Calibration of the parameters affecting the financial accelerator follow BGG (999): the monitoring cost, µ, expressed as a proportion of entrepreneurs total gross revenue, is.. Default rate of entrepreneurs is 3 percent per year, and the variance of the idiosyncratic productivity to entrepreneurs is.8 The share of total government spending of GDP is set equal to 3 percent. The government debt to GDP ratio is. The steady state private capital income tax rate, τ K, is set to.5, while τ C =.. Lump-sum transfers τ equals.6. Given these choices, the government s intertemporal budget constraint implies that τ N equals.33 in steady state. Same paths for τ C,t and G I,t as in stylized model Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 5 / 3
Impulse Responses to Gradual Sales Tax Hike percent of baseline GDP hike in fully-fledged model Figure 4. Impulses to Sales Taxes in Normal Times and in a Quarter Liquidity Trap in the Full Model.5 Non-agressive Tax Rule.5 Agressive Tax Rule Potential Normal Times Liquidity Trap.5.5 -.5 -.5 Real Interest Rate (APR) Real Interest Rate (APR).5.5.5.5 -.5 -.5 - - Labor Income Tax Labor Income Tax - - - - -3-3 Govt. Debt (share of trend GDP) Govt. Debt (share of trend GDP) - - - - -3-3 -4-4.4.4.. Quarter Quarter Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 6 / 3
Impulse Responses to Gradual Sales Tax Hike What drives the difference w.r.t. stylized model, focus on real wage Figure 6. Simulations of Impulses to Sales Tax in Simplified and Full Models Sticky Price Model Real Wage 6 Potential 4 Normal Times Liquidity Trap - Sticky Price - Sticky Wage Model Real Wage.5.5 -.5 - - Sticky Price - Sticky Wage Model with Habit Consumption Real Wage.5.5 -.5 - - Full Model Real Wage.5.5 -.5 - - Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 7 / 3
Impulse Responses to Gradual Sales Tax Hike What if capital taxes are used aggressively to stabilize government debt? Figure 6. Impulses to Sales Taxes with Aggr. Rule on Cap. Income Tax in the Full Model 4 Real Interest Rate (APR) 3 Potential.5 -.5 Normal Liquidity Trap - Total Consumption 4 HtM Consumption - - - Optimizing Households Consumption - - - Capital Income Tax (P.P.) - -3 Government Debt (Trend Share).5 -.5 5 5 5 Private Investment -.5 Policy Rate (APR).5.5 -.5 Quarter Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 8 / 3
Impulse Responses to Higher Government Investment percent of baseline GDP hike in fully-fledged model Figure 7. Impulses to Public Invest. in Normal Times and in a Quarter Liqu. Trap in the Full Model Non-agressive Tax Rule Agressive Tax Rule Potential Normal Times Liquidity Trap - - Real Interest Rate (APR) Real Interest Rate (APR) 3 3 - Labor Income Tax - Labor Income Tax 6 6 4 4 6 Govt. Debt (share of trend GDP) 6 Govt. Debt (share of trend GDP) 4 4.5.5 Private Capital Stock Private Capital Stock - - Quarter Quarter Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 9 / 3
Impulse Responses to Higher Government Investment Robustness to alternative assumptions for non-aggressive rule in fully-fledged model Figure 8. Alternative Simulations of Impulses to Public Invest. in the Full Model Benchmark calibration Potential Normal Times.5 Liquidity Trap - - -.5 - Public infrast. is not productive.5 -.5 - Public infrast. is more productive.5 -.5 - - - Public infrast. is sooner productive (- years).5 -.5 - Public infrast. is later productive (5- years).5 -.5 - - Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 3 / 3
Smaller Effect of UFP in a Stylized Model with GHH pref. responses in models with GHH preferences and non-agg. tax rules Stylized Model with GHH preferences, Non-agg. Tax Rule Sales Tax Shock Gov t investment shock Workhorse Model with GHH preferences, Non-agg. Tax Rule Sales Tax Shock Gov t investment shock Lemoine and Lindé (BdF and Riksbank) Conventional vs Unconventional Fiscal Policy ESCB Cluster, /8 3 / 3