Monetary Policy Surprises, Credit Costs. and. Economic Activity

Transcription

1 Monetary Policy Surprises, Credit Costs and Economic Activity Mark Gertler and Peter Karadi NYU and ECB BIS, March 215 The views expressed are those of the authors and do not necessarily reflect the offi cial views of the ECB or the Eurosystem

2 Conventional Monetary Policy Transmission 1. Aggregate spending depends on current and expected future real interest rates 2. Central bank controls nominal short rate i t 3. Nominal rigidities imply control over current and expected future short real rates, at least for some horizon. 4. Expectations hypothesis ) policy transmitted via yield curve Loglinear approx of m period zero-coupon gov t bond ) i m m 1 1 t = E t m f X j= i t+j g + m t m t term premium 1

3 Two Elaborations 1. Forward Guidance: (a) CB a ects yield curve by managing expectations of future path of i t : 2. Credit Channel(e.g., Bernanke and Gertler 1995) (a) With credit market frictions, to a rst order i b t = i t + x t i b t private borrowing rate; x t external nance premium (credit spread) (b) Monetary policy a ects credit spreads as well as risk free rate i. i t " ) tightening of credit frictions ) x t " ii. vice versa for i t # 2

4 What We Do Analyze joint response of economic activity and various credit cost measures to "exogenous" monetary policy surprises =) To do so, we combine: Traditional "money shock" VAR analysis (e.g. BB 1991, CEE 1996) High frequency identi cation (HFI) of policy surprises on interest rates (e.g, Kuttner, 21, GSS, 25) Policy surprises: Unexpected changes in interest rate futures on FOMC dates Use HFI measures of policy surprises as "external instruments" in monthly VARs: 3

5 Why We Do It This Way Two problems with identi cation of policy shocks in standard VARs Simultaneity between policy indicator and other nancial variables Measure of policy shocks do not incorporate shocks to foward guidance: HFI addresses both simultaneity and forward guidance issues Policy shocks are surprises in interest rate futures on FOMC dates Dependent variables are same-day responses of various asset returns. Permits incorporating use of forward guidance in policy action (GSS 25) Innovation in non-current futures rates re ects revision in beliefs about future path of rates 4

6 Why We Do It This Way (cont) Limitations to HFI With daily data di cult to identify the persistence of the e ects of policy shocks on nancial variables Can t identify joint response with economic activity Our approach: combines strengths of VAR and HFI methodologies By using futures rate surprises as external instruments, exploits HFI approach to identify exogenous policy surprises Uses VAR to trace out full dynamic response of real and nancial variables. 5

7 Preview of Main Findings 1. FF futures surprises ) responses in output and in ation consistent conventional monetary transmission mechanism and with existing VAR literature. 2. "Modest" movements in short rates ) "large" movements in real credit costs (a) Due mainly to reaction of term premia and credit spreads (b) Evidence against baseline model of monetary policy transmission. i. Still evidence for sticky prices: real rates move one for one with nominal ii. Need to adjust model to account for term premia and credit spread responses. 3. Forward guidance important to strength of policy transmission. 6

8 Methodology VAR with external instruments (Stock-Watson (212), Mertens-Ravn 213). Structural autoregressive model Reduced form model with B j = A 1 C j ; S = A 1 AY t = Y t = px j=1 px j=1 C j Y t B j Y t u t = S" t j + " t j + u t 7

9 Methodology (cont ) y p t Y t monetary policy indicator; " p t structural policy shock s column in S corresponding to impact on each element of u t of " p t To compute the impulse response to a monetary shock, need to estimate Y t = px j=1 B j Y t j + s" p t B j obtained via least squares; need restrictions to identify s Standard restriction: elements of s are zero except the one corresponding to the reduced form residual for the policy instrument. 8

10 External Instruments Z t vector of instrumental variables " p t vector of structural shocks not including policy shock Z t must satisfy E[" p t Z t] = E[" p t Z t ] = 9

11 External Instruments (con t) u p t u q t reduced form residual from equation for policy indicator reduced form residual for variable q 6= p: s q " p t response of uq t to "p t : Goal: Identify s q which gives responses of u q t to the policy shock "p t Use 2SLS: Three steps: Obtain u t from OLS regression of reduced form VAR To identify variation in u p t due to "p t ;regress up t on Z t To obtain estimates of s q ; regress u q t on up t ; using the tted values c u p t from the rst stage regressions as instruments for u p t. 1

12 Daily Fed Funds Futures Surprises as Instruments f t+j settlement price on FOMC day in month t for FF futures expiring in t + j f t+i; 1 settlement price on day prior to FOMC meeting i u t+i surprise in target rate expected for month t + j on FOMC day in month t : i u t+j = f t+j f t+j; 1 i u t shock to current funds rate target (Kuttner, 211) for j 1; i u t+j shock to target expected at t + j: (GSS, 25) i u t+j measured within 3 minute window of FOMC decision Isolates FOMC news (GSS, 25). 11

13 Policy Indicator (vs. Policy Instrument) Monthly VARs with IP, CPI, various interest rates and a policy indicator Policy indicator ( i.e., the "policy relevant" interest rate) Re ects stance of monetary policy, encompassing forward guidance. Residual incorporates policy shocks, including shocks to forward guidance Conceptually preferred indiciator: two year government bond rate View that FOMC operates with 2 yr horizon for Funds rate, (e.g. Swanson- Williams, 212, Hanson-Stein, 212) We use one year government bond rate as policy indicator for pragmatic reasons Avoids potential weak instruments problem Results robust to using two year rate as a policy indicator 12

14 Policy Indicator and Exogenous Policy Surprises Given monthly frequency, return on 1yr govt bond rate i 12 t i 12 t = 1 12 E tf Reduced form VAR residual for i 12 t i 12 t E t 1 i 12 t = X j= 11X j= i t+j g + 12 t equivalent to: fe t i t+j E t 1 i t+j g + 12 t E t 1 12 t Instrumenting with FF, ED rate surprises isolates orthogonal movements i.e, Isolates orthogonal surprises in current and expected future short rates. Policy shock is linear combination of surprises in di erent FF and ED futures 13

15 Data Description Sample: 1979:9-212:6 Economic variables: IP, CPI Interest rates Gov t bond yields: 1yr (policy indicator), 2yr, 5 yr, 1 yr; 1m FF rate Baa spread, Gilchrist/Zakrasjek excess bond premium Mortgage.spread, commercial paper spread Instruments: available 1991:1 through 212:6 1m, 3m ahead FF futures; 6m, 9m, year ahead 3 month ED futures We use 3m ahead FF futures as baseline (best instrument choice) Results robust to other instrument combinations 14

16 Figure 1: 1 year rate shock with excess bond premium 1 year rate 1 year rate.4.2 External Instruments.4.2 Cholesky CPI.2 CPI IP IP Excess Bond Premium.2 Excess Bond Premium First stage regression: F: robust F: R2: 7.76 Adjusted R2: 7.4

17 Figure 2: 1 year rate shock with corporate and mortgage premia.6 1 year rate.1 CPI IP.15 Excess Bond Premium Mortgage spread Commercial Paper spread (3 months) First stage regression: F: robust F: R2: 7.78 Adjusted R2: 7.42

18 Figure 3: 1 year rate shock: Response of other interest rates Federal Funds Rate 1 year rate year rate year rate year rate x5 forward First stage regression: F: robust F: R2: 7.78 Adjusted R2: 7.42

19 Figure 4: 1 year rate shock: Response of real rates and breakeven inflation rates year rate Real Nominal year breakeven inflation rate year rate.1 5 year breakeven inflation rate year rate.1 1 year breakeven inflation rate First stage regression: F: robust F: R2: 7.78 Adjusted R2: 7.42

20 Calculating Term Premia and Excess Return Responses Term premium on m period gov t bond, m t : m t = i m t Obtain response of i m t and i t from VAR Use path of i t to compute E t f m P 1 j= Excess return on private m period bond, t m 1 m E X1 tf i t+j g j= i t+j g for each t: i mp t t = i mp t = (i mp t rate on m period private bond m 1 m E X1 tf i t+j g j= i m t ) + m t 15

21 Figure 5: 1 year rate shock: Response of term premia and excess premia 1 year rates 2 year rates.2 Term premium Nominal rates year rates 1 year rates Corporate bond Mortgage.2 Combined excess premium Nominal rates First stage regression: F: robust F: R2: 7.78 Adjusted R2: 7.42

22 Figure 7: Federal Funds rate shock 1 year rate CPI.2 1 year FF IP Excess Bond Premium Mortgage spread Commercial Paper spread (3 months) year rate.2 1 year rate First stage regression: F: robust F: 14.4 R2: 8.95 Adjusted R2: 8.59

23 Table 4: Effects of private information on tight window monetary policy surprise ( ) (1) (2) (3) VARIABLES FF1 FF4 ED4 π.254*.195**.29 (1.818) (2.126) (1.632) dy **.299** (1.418) (2.39) (2.511) π -.435** -.342*** (-2.588) (-2.861) (-1.591) dy (-.449) (-.741) (-1.39) Observations R-squared F-statistic prob > F Robust t-statistics in parentheses *** p<.1, ** p<.5, * p<.1

24 Figure 1: 1 year rate shock with instruments without the Fed s private information, year rate CPI.2 FF4 FF4, No Private Info IP.2 Excess Bond Premium Mortgage spread Commercial Paper spread (3 months) year rate 1 year rate First stage regression: F: 5.25 robust F: 6.77 R2: 2.53 Adjusted R2: 2.5

25 A Model Consistent with Facts: Gertler-Karadi 212 Baseline: conventional monetary DSGE (CEE 25) Banks intermediate funding of private securities and govt bonds Financial frictions introduce balance sheet constraints on banks ) Limits to arbitrage that depend inversely on balance sheet strength Frictions greater for private securities than for gov t bonds Contractionary monetary policy shock increases both term premia and credit spreads Tightening weakens bank balance sheets ) Tightens limits to arbitrage, raising term premia and credit spreads Ampli es impact on economy. 17

26 Concluding Remarks VAR with FF/ED futures as external instruments used to study monetary policy transmission Key ndings: Responses of output and in ation consistent with earlier VAR analysis "Modest" movements in short rates ) "large" movements in credit costs Due to responses of term premia and credit spreads Forward guidance enhances impact of policy Main implication: need to modify conventional model to allow for term premia and credit spread e ects. 18

27 Figure 8: 2 year rate shock with a full set of GSS instruments 1 year rate CPI.2 1 year, FF4 2 year, GSS IP Excess Bond Premium Mortgage spread Commercial Paper spread (3 months) year rate.2 1 year rate First stage regression: F: 2.65 robust F: 3.99 R2: 4.99 Adjusted R2: 3.1

28 Figure 9: 1 year rate shock, year rate CPI IP Excess Bond Premium Mortgage spread Commercial Paper spread (3 months) year rate.2 1 year rate First stage regression: F: robust F: R2: 7.81 Adjusted R2: 7.37

29 Table 1: Yield effects of monetary policy shocks (event study, daily, ) Indicator & (1) (2) (3) (4) (5) (6) (7) Instruments 2 yr 5yr 1yr 3yr 5x5 forw baa + Mortg. + FF,FF1.367***.233** (3.467) (2.241) (1.53) (.13) (-.388) (1.475) (1.445) 1YR,FF1.739***.469*** (8.493) (3.94) (1.173) (.13) (-.379) (1.544) (1.416) 1YR,FF4.88***.683***.375***.145* **.427** (15.81) (8.21) (4.41) (1.694) (.614) (2.176) (2.239) 2YR, FF4.778***.432***.169* **.483** (11.8) (5.36) (1.839) (.72) (1.986) (2.141) 2YR, GSS.878***.575***.234***.271***.231*.35** (18.7) (11.84) (4.139) (3.61) (1.844) (2.49) Robust z-statistics in parentheses *** p<.1, ** p<.5, * p<.1 QE dates and crisis period are excluded, 188 observations + : 2-week cumulative changes

30 Table 2: TIPS and breakeven inflation effects of monetary policy shocks (daily event study, ) Indicator & (1) (2) (3) (4) (5) (6) Instruments TIPS 2yr TIPS 5yr TIPS 1yr Bkeven 2yr Bkeven 5yr Bkeven 1yr FF, FF **.149** ** (1.348) (2.217) (2.287) (.596) (-1.553) (-2.81) 1YR, FF1.8***.639***.384***.282* (4.141) (7.66) (6.121) (1.913) (-.62) (-1.165) 1YR, FF4.84***.565***.315*** (5.171) (5.763) (4.136) (.474) (.269) (-.815) 2YR, FF4.759***.618***.344*** (5.9) (4.32) (3.592) (.525) (.269) (-.743) 2YR, GSS.754***.63***.462***.196**.189**.11* (7.749) (8.394) (9.35) (1.981) (2.165) (1.818) Robust z-statistics in parentheses *** p<.1, ** p<.5, * p<.1 QE dates and crisis period are excluded, 58 (2yr), 1 observations + : 2-week cumulative changes

31 Table 3: Effects of high-frequency instruments on the first stage residuals of the 4 variable VAR (monthly, ) (1) (2) (3) (4) (5) (6) (7) (8) (9) (1) VARIABLES 1YR 1YR 1YR 1YR 1YR 2YR 2YR 2YR 2YR 2YR FF1.89*** **.174 (4.44) (1.129) (2.116) (.462) FF *** 1.266*** 1.243***.779** 1.13*** 1.379*** (4.184) (4.224) (3.68) (2.272) (2.643) (3.361) ED (1.244) (.859) ED *** ** (-2.635) (-2.448) ED4.624** ** ** (2.39) (-.476) (2.493) (.923) (-.863) (2.465) Observations R-squared F-statistic Robust t-statistics in parentheses *** p<.1, ** p<.5, * p<.1

32 Term Premia Responses Using Expectations Data Term premia responses may re ect "non-rational" forecasts of future short rates Can evaluate using survey data on expectations: Blue Chip Economic Indicators survey: 3 month T-bill rate forecasts up to 6 quarters ahead; available 1983:3-212:6. Results: At longer horizons (5-1 years): Cannot reject that market expectations of the future path of the Funds rate are "rational"; i.e. consistent with the impulse responses of the Funds rate. Term premium e ects not due to "irrational expectations." As shorter horizons (1-2) years "Over-reaction" of expectations could explain term premium e ects Data is too noisy to say for sure. 16

33 Figure 6: 1 year rate shock: Response of private sector expectations, year rate (FF) Expectations Nominal Implied riskless rates year rate (FF) year rate (FF).2 1 year rate (FF) First stage regression: F: robust F: R2: 7.78 Adjusted R2: 7.42