Monetary Policy Surprises, Credit Costs, and Economic Activity

Questions & answers

Q1. What is the paper’s central research question, and what identification strategy does it use?

The paper asks whether “modest” monetary-policy-induced movements in short-term interest rates produce commensurately modest movements in private credit costs, as the standard frictionless transmission mechanism would predict, or whether term premia and credit spreads amplify the effect into economically large credit-cost movements. Identification uses “external instruments” (a proxy SVAR, following Stock and Watson 2012 and Mertens and Ravn 2013) rather than the timing/ordering restrictions of a conventional recursive VAR: surprises in federal funds and Eurodollar futures prices in a narrow window around FOMC announcements serve as instruments that are relevant (correlated with the true structural policy shock) but exogenous (uncorrelated with other structural shocks), with the structural shock’s impact effects recovered via two-stage least squares.

Q2. Why does the paper use the one-year government bond rate, rather than the federal funds rate, as its baseline policy indicator?

The one-year rate is chosen because it captures revisions in beliefs about the expected future path of short-term rates — including forward guidance effects — whereas the federal funds rate captures mainly surprises in the current, very short rate; the authors describe the two-year rate as “conceptually preferred” but do not use it as the baseline because interest rate futures surprises are not strong instruments for monthly VAR innovations in the two-year rate (first-stage F-statistics fall below the conventional threshold of 10 for all instrument combinations they try). The baseline instrument (a three-month-ahead federal funds futures surprise, FF4) produces strong first-stage statistics for the one-year rate (F-statistics around 21-22, well above the weak-instrument threshold) in both the paper’s simple four-variable and baseline six-variable VAR specifications.

Q3. What are the main real and price-level effects of a monetary policy surprise?

A one-standard-deviation contractionary surprise — inducing roughly a 20-25 basis point rise in the one-year rate — produces a significant and fairly rapid decline in industrial production, with the trough reached after roughly 18 months in the baseline specification, while the decline in the CPI is not statistically significant. Consistent with a high-frequency-identification exercise (Table 2), virtually all of the policy surprise’s impact falls on real rather than nominal interest rates — real and nominal rates move “virtually one-for-one” — and breakeven (market-implied) inflation expectations show only a small, marginally significant decline at five- and ten-year horizons, otherwise remaining statistically insignificant.

Q4. What is the paper’s central finding about the amplification of credit costs?

For a roughly 20 basis-point rise in the one-year rate, the corporate bond rate rises by roughly 15 basis points and the mortgage rate by roughly 7 basis points — a large amplification relative to the size of the underlying policy surprise — and the Gilchrist-Zakrajšek excess bond premium rises by 8 basis points on impact and stays elevated for roughly eight months, while the three-month commercial paper spread rises about 5 basis points for four to five months. The authors are explicit that this amplification is “due mainly to the reaction of both term premia and credit spreads,” not to any commensurately large revision in the expected future path of short-term policy rates.

Q5. How does the paper decompose the response of longer-term interest rates into expected-rate, term-premium, and credit-spread components?

Decomposing the response using VAR-implied expectations of future short rates (cross-checked against Blue Chip survey forecasts), the paper finds that for the two-, five-, and ten-year maturities, “virtually all the rate increase is due to the term premium, with no impact of the path of expected short rates,” while for the one-year rate itself, roughly 80% of the movement is attributable to a term-premium effect. Across maturities, “virtually all the movement in rates is due to the excess premium” — defined as the sum of the credit spread and the term premium — reflecting the fact that the underlying policy shock’s effect on the expected path of short rates is itself fairly transitory, leaving little residual expected-rate contribution to longer-maturity yields.

Q6. Why do the authors interpret these results as inconsistent with the standard frictionless transmission mechanism, and what alternative mechanism do they propose?

Under the standard frictionless view, private borrowing rates should move essentially one-for-one with revisions in the expected path of the policy rate, with no independent response of term premia or credit spreads; the paper’s finding that term premia and credit spreads instead account for nearly all of the credit-cost response is, in the authors’ words, inconsistent with this benchmark. They interpret the credit-spread component as consistent with a credit channel in the spirit of Bernanke and Gertler (1995) — financial frictions raise the external finance premium when policy tightens — while the additional term-premium response points to a further friction, such as limited participation in or limits to arbitrage across bond markets of different maturities, and they cite their own companion structural model (Gertler and Karadi 2013) — a sticky-price monetary DSGE extended with limited financial-market participation — as one framework capable of generating both effects simultaneously.

Q7. What does the paper find about the importance of forward guidance relative to surprises in the current policy rate?

Substituting the federal funds rate for the one-year rate as the policy indicator (using the contemporaneous funds-futures surprise, FF1, as the instrument) and normalizing to produce the same size funds-rate movement as the baseline, the resulting output contraction is “more than 50 percent smaller” than in the baseline, and the CPI decline and the responses of credit spreads and long-term rates are similarly weaker. This is offered as direct evidence that a policy surprise which revises market expectations about the future path of short rates (forward guidance) has materially stronger real and financial effects than an equally-sized surprise confined to the current policy rate alone — supporting the paper’s choice of the (partly forward-guidance-sensitive) one-year rate as the more informative policy indicator.

Q8. What robustness checks and limitations do the authors discuss regarding instrument validity?

Because Romer and Romer (2000) raise the concern that Fed private information about the economic outlook could contaminate interest-rate futures surprises (violating instrument exogeneity), the authors construct a “purified” instrument that strips out variation explained by proxies for Fed private information (which account for only about 15% of the baseline instrument’s variance) and find the estimated effects on output, inflation, and spreads become larger, not smaller, when this private-information component is removed — though this purified-instrument specification has notably weaker first-stage statistics, so the authors caution that “weak instruments concern cannot be ruled out” for this particular check. They also show the main results are robust to using the full Gürkaynak-Sack-Swanson instrument set, to a pre-crisis-only sample (1979-2008), and to excluding the 2008-2009 financial crisis period from the instrument-relevance regressions, though they flag that the term-premium estimates rely on the VAR’s own implied measure of interest-rate expectations, which — even when partially validated against Blue Chip survey forecasts — carries enough uncertainty that “it is very hard to reach definitive conclusions” about whether market overreaction contributes to the estimated term-premium response at short maturities.