Financial Factors in Economic Fluctuations Lawrence Christiano Roberto Motto Massimo Rostagno What we do • IIntegrate t t financial fi i l frictions f i ti into i t a standard t d d equilibrium ilib i model and estimate the model using Euro Area and US data. – Asymmetric information and costly state verification (Townsend (1978), Bernanke-Gertler-Gilchrist (1999)) – Endogenous determination of financial liabilities, like M1 and M3 (Chari-Christiano-Eichenbaum (1995)) • Decompose 14 aggregate data series into shocks and propagation mechanisms: – A new shock, a ‘risk’ shock – A new source of propagation: non-state contingent nominal rates of interest. Outline • Sketch the basic ingredients of the model. • Results Standard Model Firms consumption Yt 0 Yjt 1 1 f,t f,f,t dj Investment goods , 1 ≤ f,t , Yjt t Kjt z t l jt 1− Supply labor It ̄ other oil t au t Kt G t C t ,t ≤ Y t . Rent capital Households Backyard capital accumulation: u c,t R kt1 E t c,t u c,t1 t1 ̄ t1 1 − K ̄ t G i,t , I t , I t−1 K R kt1 u t11 r kt1 1 − P k ′,t1 P k ′,t Financial Sector • Assets: A t – Working capital loans – Loans to entrepreneurs for purchasing capital • Liabilities: – Provide utility to households – Neoclassical bank production function • Results: – features on the liability side of banks do not add shocks or propagation to business cycle dynamics – Features on the asset side do matter…particularly ‘loans to entrepreneurs’ Entrepreneur • Buys capital from capital producers using own resources and loans from bank • Rents capital to intermediate good producers. d Banks Households Banks, Households, Entrepreneurs ~F, t , E 1 entrepreneur entrepreneur entrepreneur Households Bank entrepreneur entrepreneur Standard debt contract Economic Impact of Risk Shock llognormall di distribution: t ib ti 20 percent jump in standard deviation 1 0.9 σ σ *1.2 0.8 Larger number of entrepreneurs in left tail problem for bank density 0.7 0.6 Banks must raise interest rate on entrepreneur 0.5 Entrepreneur borrows less 0.4 0.3 Entrepreneur buys less capital, investment drops, economy tanks 0.2 01 0.1 0.5 1 1.5 2 2.5 idiosyncratic shock 3 3.5 4 Risk Shocks and News • Finding: – Economic effects of risk shocks operate primarily via advance information, or news. Risk Shock and News • Assume iid Wold innovation in log t iid, log t log t−1 ut • A Agents t have h advance d iinformation f ti about b t pieces of u t u t 0t 1t−1 . . . 8t−8 2 it−i ~iid, E it−i 2i it−i ~piece of u t observed at time t − i Monetary Policy • Nominal rate of interest function of: – Anticipated A ti i t d llevell off iinflation fl ti and d change. h – Slowly moving inflation target. – Deviation off output growth from f ss path. – Growth of credit in case of EA. – Monetary policy shock. Estimation • EA and US data covering 1985Q1-2008Q2 Δ log per capita stock market indext Pt GDP deflator inflationt logper capita hourst per capita credit t Pt Δ log Δ logper capita GDP t Δ log Hourly compensationt Pt Δ logper capita investment t Xt Δ log per capita M1 t Pt Δ log per capita M3 t Pt Δ logper capita consumptiont Risk Spread t R long − R et t R et Δ logPIt ) Δ logreal oil price t Δ log per capita Bank Reserves t Pt • Standard Bayesian methods Summary of results • Risk shocks: – important source of fluctuations. fluctuations – news on the risk shock important. • Assumption of state non non-contingent contingent interest rates has an important impact on shock propagation. • Money M demand d d and d iinside id money: – relatively unimportant as a source of shocks – modest contribution to forecast ability • Out-of-Sample Out of Sample evidence suggests the model deserves to be taken seriously. Risk Shocks • Important • Why Wh are th they iimportant? t t? • What shock do they displace, and why? Figure: Year-over-year GDP Growth Rate - Data (black) versus what data would have been with only the risk shock US Risk shock important EA Variance Decomposition of Financial Shocks, US Data risk shock, t Output IImportant t t for f output and investment 19 Investment 38 Consumption 5 Very important for financial variables Risk Spread 97 R l Value Real V l off Stock St k Market M k t 80 Although the data like signals on standard technology shocks they prefer signals on risk shocks, risk. Why y Risk Shock is so Important p • A. Our econometric estimator ‘thinks’ risk spread p ~ risk shock. • B B. In the data: the risk spread is strongly negatively correlated with output. • C. In the model: bad risk shock generates a response that resembles a recession. recession • A+B+C suggests risk shock important important. Correlation (risk spread(t),output(t-j)), HP filtered data, 95% Confidence Interval The risk spread is significantly negatively correlated with output and leads a little. Notes: Risk spread is measured by the difference between the yield on the lowest rated corporate bond (Baa) and the highest rated corporate bond (Aaa). Bond data were obtained from the St. Louis Fed website. GDP data were obtained from Balke and Gordon (1986). Filtered output data were scaled so that their standard deviation coincide with that of the spread data. Another Reason the Risk Shock is so Important • Positive shock to risk triggers what looks like a recession US, Impulse Response to Riskiness Shock (contemporaneous) Credit procyclical, procyclical risk spread countercyclical Output Investment Consumption 0 0.05 0.2 -0.1 01 0 percent p percent p -0.05 -0.2 -0.05 -0.4 -0.15 -0.1 -0.6 Risk spread (AR) Premium (Annual Rate) Real Net Worth 0 Total Loans (Real) -0 0.5 5 -1 -2 percent 100 basis points b percent percent 0 50 -1 -1.5 -2 -3 -2.5 0 10 20 30 10 20 30 10 20 30 What Shock Does the Risk Shock Displace, and why? • Th The risk i k shock h k crowds d outt some off th the role of the marginal efficiency of i investment t t shock. h k Business Cycle Frequencies Model risk shock, t survival, t marginal efficiency of investment, it Output Baseline 19 2 22 CEE-SW CEE SW na na 46 Investment Baseline 38 5.2 53 CEE-SW na na 91 Consumption Baseline 5 1 2 CEE-SW na na 3 ExternaltoFinance Premium Risk shock appears crowd out the marginal 9 efficiency 3of investment shock 97 0 Baseline li CEE-SW na na na Real Value of Stock Market Baseline 80 10 2 CEE-SW na na na Why y does Risk Crowd out Marginal Efficiency of I Investment? t t? Price of capital Supply shifter: marginal efficiency of investment, i,t Demand shifters: risk shock, t ; wealth shock, shock t Quantity of capital • Marginal efficiency of investment shock can account well for the surge in investment and output in the 1990s 1990s, as long as the stock market is not included in the analysis. analysis • Wh When the th stock t k market k t iis iincluded, l d d th then explanatory power shifts to financial market k t shocks. h k Variance Decomposition of Inside Money Shocks Business Cycle y Frequencies q Bank demand for excess reserves, t Bank technology shock, x bt Household money demand shock, t Output 0 05 0.5 0 Investment 0 0 0 Consumption 0 1 0 Risk spread 0 0 0 Real value of stock market 0 0 Contribution to variance: trivial 0 Out of Sample RMSEs • There is not a loss of forecasting power with the additional complications of the model. • The model does well on everything, except the risk spread. • Calculations for 1,2,…,12 ahead forecasts: – First date of forecast, forecast 2001Q3 – DSGE models re-estimated every other quarter – BVAR re-estimated each quarter (standard Minnesota priors). Other support for the model • Model predicted default rates positively correlated with measures of default in the data. 0.0235 4 Default Probability (lhs) Expected Default Probability, NFC (rhs) 0.0185 2 0.0135 0.0085 1990Q1 0 1993Q1 1996Q1 1999Q1 2002Q1 2005Q1 2008Q1 Our Measure of Idiosyncratic Risk versus Bloom, Risk, Bloom et al 1.15 1.05 0.5 Series1 Series4 Ri ki Riskiness Sh Shockk (3-quarter (3 t MA) Cross-firms Sale Growth Spread (3-quarter MA) 04 0.4 0.95 0.85 0.3 0.75 0.65 1981Q2 0.2 1986Q2 1991Q2 1996Q2 2001Q2 2006Q2 Conclusion • Incorporating financial frictions changes inference about the sources of shocks and of propagation – risk shock shock. – New nominal rigidity: nominal non-state contingent interest rate • Models with financial frictions can be used to ask i t interesting ti policy li questions: ti – When there is an increase in risk spreads, how policy y respond? p should monetaryy p – How should monetary policy be structured to avoid excess asset market volatility? – What are the pro’s pro s and con’s con s of ‘unconventional unconventional monetary policy’?