1. Credit, Risk Appetite,
and Monetary Policy
Transmission
D AV I D A I K M A N , A N D R EA S L E H N E RT, N E L L I E L I A N G , M I C H E L E M O D U G N O
J U N E 5 , 2 0 1 7
V I E W S E X P R E S S E D A R E O U R O W N A N D N O T N E C E S S A R I L Y T H E V I E W S O F T H E F E D E R A L R E S E R V E B O A R D , B A N K O F E N G L A N D , O R S T A F F

2. Motivation
The global financial crisis highlighted the potential role of financial factors for the real economy
Long tradition linking risk appetite to business fluctuations (Keynes “animal spirits”)
High credit and asset valuations predict subpar economic performance (Borio and Lowe, 2002;
Drehmann and Juselius, 2015; Schularick and Taylor, 2012)
High credit growth and asset bubbles lead to weaker economic recoveries (Jorda, et al 2013)
Credit is a commonly cited financial imbalance: how do macroeconomic dynamics change if it is
elevated?
◦ Response to risk appetite shocks
◦ Monetary policy transmission

3. Outline
We characterize the time series of the credit-to-GDP gap and “risk appetite,” 1975 to 2014
We estimate VAR models of the macroeconomy and monetary policy
◦ Augmented with our risk appetite measure and the credit-to-GDP gap
◦ Threshold VAR allows for nonlinear dynamics
We characterize the response to
◦ Risk appetite shock
◦ Monetary policy shock
We split the sample into periods when the credit-to-GDP gap is high or low to test for
nonlinearities

4. Key empirical results
Our risk appetite measure (“ALLM”)
◦ Is an indicator of financial conditions and is expansionary
◦ But it can lead to a higher credit-to-GDP gap and recession
Dynamics are nonlinear depending on nonfinancial credit-to-GDP gap. When gap is high:
◦ ALLM shocks lead to recessions
◦ Monetary policy effect is attenuated
When the credit gap is high, monetary policy:
◦ Does not affect GDP, unemployment or inflation
◦ Does not affect our risk appetite measure
◦ Using Hanson-Stein (2015) framework, less transmission to yields 5 to 9 years out
Policy attenuation result also holds when the credit gap is growing (can’t stop the boom)

5. Recent papers on similar topics
Alpanda and Zubairy (March 2017)
◦ Monetary policy transmission varies with level of household debt
Ottonello and Winberry (May 2017)
◦ The level and distribution of business debt affects monetary policy transmission
◦ Rates ↓ → more indebted firms pay down debt, less indebted firms increase investment
Brunnermeier, Palia, Sastry, and Sims (April 2017)
◦ 10 variable VAR identified using Rigobon (2003) heteroskedasticity strategy – some periods see more
volatile shocks
◦ Includes HH & business credit, GZ, spreads, monetary policy, real activity
◦ Shocks that matter: monetary policy, financial stress
◦ Shocks that don’t matter: credit (household or business)
◦ Business credit does matter in a small system – y, p, BC, HHC

6. ALLM is in the tradition of financial
conditions indexes (FCIs)
•FCIs attempt to measure stimulus/contraction from financial conditions – private borrowing
rates, stock prices, the exchange value of the dollar
•Monetary policy → conditions → real economy
• Magnitude, timing potentially time-varying
•Post-crisis a resurgence of interest in FCIs including financial stress indexes:
• Broad review, focus on macro forecasting performance, “neoclassical” vs “non-neoclassical” variables
(Hatzius, Hooper, Mishkin, Schoenholz, and Watson 2010).
• A number of FCIs developed and routinely updated: Aramonte, Rosen, Schindler 2013 evaluate 12
separate indexes and evaluate them as early warning indicators and coincident indicator
• Some indexes rooted in theory, e.g. Gilchrist and Zakrajsek (2012)’s excess bond premium, which uses
micro data on credit spreads to measure the residual after controlling for default risk

7. Constructing ALLM: Variables related to
lenders’ willingness to make riskier loans
Thought exercise: Want variables that measure lenders’ appetite for risk in making loans to
households and (nonfinancial) businesses (including commercial real estate)
1. Equity markets: stock market volatility and the S&P 500 price-earnings ratio.
2. Business credit: Triple-B corporate bond spread to Treasury, the share of nonfinancial
corporate bond issuance that is speculative-grade, and the index of credit availability from
the NFIB survey of small businesses.
3. Commercial real estate: a commercial real estate price index deflated into real terms and
commercial real estate debt growth.
4. Household: the residential price-to-rent ratio and lending standards for consumer
installment loans from the Senior Loan Officer Opinion Survey (SLOOS).

8. ALLM v1.0 and v2.0
ALLM v1.0 contains asset prices (VIX, P/E) and sentiment variables
So shocks to ALLM v1.0 could be:
◦ Valuation shocks: Investor risk sentiment or appetite (separate from financial accelerator effects)
◦ Lending standards shocks: Profitability of intermediaries (He and Krishnamurthy (2012, 2013) and
Gilchrist and Zakrajsek (2012))
Ongoing work exploring separating valuation and lending standards terms
◦ Identification is cleaner
Showing you results from ALLM v1.0 today

9. VAR specification
U.S. macro data 1975:Q1 to 2014:Q4
Log real GDP, GDP deflator, Unemployment rate
Credit-to-GDP gap
◦ Household vs. business
◦ Bank vs. nonbank
Risk appetite – asset valuations and lending standards
Federal funds rate
 We define a measure to be a vulnerability if an impulse to the measure leads to an economic
contraction

10. VAR dynamics
Shocks are identified using the Cholesky decomposition with shocks ordered as in the monetary
policy literature
◦ Monetary policy reacts to all shocks in a period
◦ ALLM reacts to all shocks within a quarter save monetary policy
◦ The unemployment rate, the GDP deflator, and real GDP react to shocks to the vulnerability measure
and monetary policy with a one-quarter lag
Estimate the VAR following Giannone, Lenza, and Primiceri (2015)
◦ Bayesian technique specifies a prior that each variable follows a random walk, possibly with a drift; this
reduces estimation uncertainty and leads to more stable inference.

11. Threshold VAR
Nonlinear estimations – often speak of financial imbalances as “high” or “low”
◦ Dynamics could differ for a variety of reasons
Effectively estimate system on disjoint sets depending on whether the credit gap is above/below its
mean
Not model transitions from one state to another
𝑦𝑡 = 𝑐 𝑗
+ 𝐴 𝐿 𝑗
𝑦𝑡−1 + 𝑢 𝑡
𝑗 𝑗 = high,if 𝐶𝑌𝑡 > 0.
𝑗 = low, if 𝐶𝑌𝑡 ≤ 0.

12. A word about the trend
When is credit “too high”?
◦ Credit-to-GDP is above its trend
How do you estimate the trend in credit-to-GDP?
HP filter 𝜆 = 400,000 – due to Borio and Lowe (2002, 2004) and Basel III
Many obvious problems
◦ But this is the canonical way to do it
◦ Undertaken lots of robustness work
◦ Any sufficiently slow-moving trend estimate is going to deliver the same results

13. Credit-to-GDP and trend

14. Credit-to-GDP gap (CY)

15. Risk appetite

16. Components of risk appetite

17. Shock to risk appetite is expansionary…

18. …even with the credit/GDP gap…

19. …but nonlinear effects: when CY is high,
leads to a recession

20. Monetary policy shock works as
expected in a linear system…

21. …but is ineffective when CY is high…

22. …and when CY is growing

23. (An ALLM v1 shock when CY is growing)

24. Attenuation by horizon (Hanson-Stein,
1975-2014)

25. Robustness tests
Alternative orderings of shocks
Credit in log level
Credit-to-potential GDP
Three states, different cutoffs for thresholds
Different ways of estimating the trend
Using growth rates instead of (detrended) levels

26. Other results (in the working paper)
Disaggregate types of credit
◦ Household vs. business – business matters a lot as in Brunnermeier et al and Ottonello & Winberry
◦ Bank vs. nonbank – banks matters more than nonbank
Alternative financial imbalances
◦ Runnable liabilities
◦ Leverage of intermediaries

27. Summary
Findings
◦ When the credit-to-GDP gap is high
◦ Economic growth subpar
◦ Economic dynamics are different – attention of monetary policy transmission
◦ Risk appetite – an indicator of financial conditions; but contributes to the buildup of credit-to-GDP
Implications
◦ Supports a story in which risk appetite shock leads → expansion & credit growth → credit bust
◦ Credit quantity, not just prices, has implications for real economic activity
◦ Macroeconomic responses are nonlinear – transmission channels may operate differently under
different conditions (Hubrich and Tetlow, 2015)
Ongoing work – revising ALLM