Taming a Minsky Cycle - with Emmanuel Farhi

Webinar Hosted from PRINCETON
 For EVERYONE, WORLDWIDE

 Taming a
 Minsky Cycle
Iván Werning with Emmanuel Farhi
MIT

11. March 2021 Markus Brunnermeier

Extrapolative Expectations & Bubbles
  Extrapolative expectations (adapted expectations in growth)
  E.g. Gennaioli & Shleifer book
  (distorted beliefs)
 Potential
  Momentum Price path

  Bubbles
 Bubble
 starts
Fundamental
innovation Fundamental
 value

 2

Good vs. bad bubbles
 New technologies and R&D investments (1998-2000)
  Overcoming QWERTY (chicken-egg) problems

 Safe Asset as a bubble (government debt < )
  Serves as precautionary savings tool
  Asset Price = E[PV(cash flows)] + E[PV(service flows)]
 dividends/interest insurance via re-trading
  2 s > 0 < 0
 Debt as Safe Asset
 Brunnermeier-Merkel-Sannikov 2020
 Real estate bubbles (2006)
 Financial innovation/liberalization bubbles
 BITCOIN
 3

Harmless vs. Dangerous Bubbles – how to fight?
 Equity financed bubbles (1998-2000)
 Credit financed bubbles (2005-2006)
  Minsky’s financing classification
  Hedge borrowing: can pay off whole debt
  Speculative borrowing: can pay off interest due
  Ponzi financing

1. Policy makers should “fight” bubbles by
 a. Leaning against during build-up
 b. Clean afterwards only
2. Policy makers should “fight” bubbles
 a. with monetary policy
 b. primarily with macro-prudential tools
 c. both

 4

Minsky’s bubble phases
 Potential
 Price path

 Bubble
 Fundamental starts
 innovation
 Fundamental
 value

 Displacement Boom Euphoria Profit-taking Panic
 phase phase phase phase

 5

Why do rational investor ride rather than attack bubbles?
 Co-opetition among rational investors Sequential awareness/learning + critical mass
  Competition: exit bubble before it bursts Kills backwards induction argument
  Cooperation: ride as long as other ride it common knowledge of bubble

 Potential
 Price path

 Bubble
 Fundamental starts
 innovation
 Fundamental
 value
 Mutual knowledge
 1. order 2. order 3. order ….

 6

Poll Questions
1. Policy makers should “fight” bubbles by
 a. Leaning against during build-up
 b. Clean afterwards only

2. Policy makers should “fight” bubbles
 a. with monetary policy
 b. primarily with macro-prudential tools
 c. both

3. Policy makers’ belief distortions and exuberance are
 a. smaller than the markets’
 b. about the same
 c. Larger than the markets’

 7

Taming a Minsky Cycle

 Emmanuel Farh
 Iván Werning

 March 202
 Markus Academy, Princeton
0

 i

Macroprudential Policy
 Macroprudential policies motivation
 nancial fragilit
 aggregate demand stabilizatio
 monetary policy constraints or dilemma
 Open economy: capital ows, dilemma

 Farhi-Werning (2013, 2014, 2016)
 Applications: capital controls, scal unions, deleveragin
 General model: pecuniary + demand externalitie
 Formula: MPCs + Wedges (Econometrica 2016

 New Today… “Taming a Minsky Cycle” (2020
 Minsky Boom Bust Cycle
 Boom: compolacency, rising asset prices and leverag
 Bust: “Minsky moment”, risk repricing, deleveragin
 Non-rational expectations, extrapolation
fi
 y

 fl
 s

 n

 fi
 …

 …

 s

 )

 )

 s

 g

 e

 g

Macroprudential

 nancial macro
 decisions impact

 e.g. credit boom e.g. low return shock
 high leverage and risk taking lower future loans
fi

Macroprudential
 macropru regulation

 nancial macro
 decisions impact

 e.g. credit boom e.g. low return shock
 high leverage and risk taking lower future loans
fi

Macroprudential
 macropru regulation

 nancial macro
 decisions impact

 e.g. credit boom e.g. low return shock
 high leverage and risk taking lower future loans
 Is there a market failure?
 Not necessarily.
 Externality needed.
fi

Macroprudential
 monetary policy?
 macropru regulation

 nancial macro
 decisions impact

 e.g. credit boom e.g. low return shock
 high leverage and risk taking lower future loans
 Is there a market failure?
 Not necessarily.
 Externality needed.
fi

Macroprudential
 monetary policy? monetary policy?
 macropru regulation

 nancial macro
 decisions impact

 e.g. credit boom e.g. low return shock
 high leverage and risk taking lower future loans
 Is there a market failure?
 Not necessarily.
 Externality needed.
fi

Macroprudential
 nancial macro
 decisions impact

 Macropru formula: linked to MPCs and wedge
 General model: incomplete markets, nancial
 constraints with prices etc. (pecuniary externalities)
fi
 fi
 s

Extrapolative Expectations
 Greenwood-Sheifer (2014): survey of investor
 expectations
Extrapolation of future stock
 is Common: Bothreturns
 Boom correlate
 & Bust with past
 returns and level of stock market

Policies to Tame a Minsky Cycle

 Elements today
 Monetary with and without macro-pr
 Macroeconomic vs. nancial stability
 Targets and instruments a la Tinberge
 trading off targets with given instrument
 assignment of targets to instrument
 Key role of endogeneity of beliefs
 …

 fi
 s

 s

 n

 u

 s

Minsky
 Unhappy with neoclassical synthesis;
 important aspects of Keynes
 but missing nancial/investmen
 too rosy on stability prospect
 Ideas
 system is endogenously unstable…
 … perfect stabilization with money and scal policy:
 impossibl
 tranquility, seeds the risk taking, that creates boom/
 bus
 nancial markets different than real economy; debt
 nancing during expansion, turns more speculative
fi
fi
 t

 …

 e

 fi

 s

 t

 fi

Minsky in
 “Stabilizing an Unstable Economy”
 Boom and role of expectation feedback
 “Instability emerges as a “unless the past is being “A rise in the price of
 period of relative tranquil validated […] none but nancial instruments or
 growth is transformed into a pathological optimists will capital assets may very well
 speculative boom […] invest.” increase the quantity
 middlemen in nance demanded […] thus breed
 change in response to the conditions conductive to
 success of the economy.” another such rise.”

 Policy implications for nancial controls…

 “We need a Theory that “External controls and “It is possible that with other
 coordinating mechanisms institutional organizations
 makes instability a
 may be needed in a and policy systems the
 normal result in our capitalist economy. Indeed, susceptibility of our economy
 economy and gives us central banking and other to nancial crises can be
 handles to control it.” nancial control devices lower than at present”
 arose as a response to the
 embarrassing incoherence of
 nancial markets.”
fi
fi
fi
 fi
 fi
 fi
 …

Related Literature
Monetary Policy: Woodford, Gali, Werning, Eggertson-
Krugman, McKay-Nakamura-Steinsson, Auclert;

Monetary Policy and Expectations: Farhi-Werning;
Angeletos-Lian; Gabaix

Macroprudential Policy: Farhi-Werning, Korinek-
Simsek, Caballero-Simsek, Bianchi-Mendoza

Extrapolative/Diagnostic Expectations: Bordalo-
Gennaioli-Shleifer, Maxted …
 …

 …

 …

Monetary

 Monetary
 Macropr

 Rational
 Expectation
+

 u

Monetary

 Monetary
 Macropr

 Rational
 IT
 Expectation
+

 u

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru
+

 u

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru

 Extrapolativ
 Expectations
+

 u

 e

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru

 Extrapolativ Lean Against
 Expectations Boom
+

 u

 e

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru

 I
 Extrapolativ Lean Against
 Expectations Boom
 Macropru
+

+

 T

 u

 e

Model Ingredients
 He-Krishnamurthy (2013) (Brunnermeier-Sannikov, 2014
 asset pricing model
 adds nominal rigidities + optimal polic
 Incomplete markets
 risky asset (Lucas tree
 risk-free short-term bon
 Two agents
 savers: save risk-fre
 borrowers
 invest in risky asse
 borrow risk-fre
 Three periods t=0,1,
 Consumption good produced 1-to-1 with labor
 Rigid wages, no in ation
:

 …

 e

 fl
 t

 e

 …

 2

 )

 d

 y

 )

Demand Determined Outpu Endowment
 (rigid wage)

 t=0 t=1 t=2

borrowin risky retur
 ZLB binds realize
& investing
 d

 g

 n

 t

Periods, States and Demographics

 Three periods t ∈ {0,1,2}
 Aggregate state ω ∈ {H, L}
 Determines dividend D2,ω of Lucas tree with
 D2,H > D2,L
 Agents i ∈ {S, B} share ϕ i

Preferences and Technology

Technolog
 t = 0,1

 t=2

Preference
 Borrower

 Savers
 s

 y

 s

Nominal Rigidities

Sticky wages normalized to on
Zero Lower Bound (ZLB) binds at t=1, not at
t=0 e

Budget Constraints

Savers

Borrowers

Labor Wedges and Output Gaps

 Labor Wedge

 Positive wedges iff negative output ga
 “Macroeconomic Stability”
 s

 p

Debt as a State Variable
 Savings of savers b1S (debt of borrowers) state
 variable at t=
 Asset price and output…
 1

Debt as a State Variable
 Savings of savers b1S (debt of borrowers) state
 variable at t=
 Asset price and output…

 Financial Fragility: two intuitions
 higher debt lower risk-taking capacity
 higher risk premia lower asset price
 lower consumptio
 higher debt higher precautionary motive
 lower natural rate lower consumptio
 Risk always key here; without it, no effect.
 1

 n

 …

 n

Value Functions and AD Externality
 S
Allocation pinned down by b1
 S S B S
Value functions V (b1 ) and V (b1 )
Aggregate demand externality if recession at t=1
(compare MRS of planner to agents’)

 Externality
 Social Marginal Utilities ≠ Private Marginal Utilities

Monetary Policy
Focus on Pareto weights that neutralize
distributive objectives (λ S /λ B = c0S /c0B)
Optimal monetary policy targeting rul

Lean against boom (μ0 < 0) iff borrowers initially
 S
levered (b0 > 0
 S
Benchmark with b0 = 0 gives standard “in ation
targeting” (IT)
 )

 e

 fl

Monetary

 Monetary
 Macropr

 Rational
 IT
 Expectation

 Extrapolativ
 Expectations
+

 u

 e

Monetary Policy an Macropru
 Optimal monetary policy targeting rul

 Macroprudential tax on borrower leverag

 Assignment of targets to instruments:
 macro stability to monetary polic
 nancial stability to macroprudential policy
fi
 y

 e

 e

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru

 Extrapolativ
 Expectations
+

 u

 e

Extrapolative Expectations
 Introduce extrapolative expectations by
 borrower

 Modeled by either
 wedge in investor Euler equation o
 subjective probabilities
s

 …

 r

AD and Belief Externality
 AD and Belief Externalit

 Belief externality reinforces AD externality as
 B,e B
 long as borrowers optimistic (c1 > c1 ) in
 equilibriu
 This will be the case.
 m

 y

Monetary Policy
Optimal Monetary Policy targeting rul

Lean against boom (μ0 < 0) if extrapolative
expectations
 e

Intuition
 “Take the punch bowl away when the party is
 still going”
 Contractionary Monetary Policy
 cools economy during boo
 cools expectations of return
 cools borrowin
 low borrowing bene cial in future

 Extrapolative expectations important

 g

 fi
 m

 s

 …

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru

 Extrapolativ Lean against
 Expectations Boom ?
+

 u

 e

Monetary+Macropru
 Optimal monetary policy again

 Macropru tax borrower leverag

 Assignment of targets to instrument
 monetary: macro stability
 nancial stability: macropru
fi

 e

 …

 s

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru

 I
 Extrapolativ Lean Against
 Expectations Boom
 Macropru
+

+

 T

 u

 e

Extrapolative Expectations During Bust

 Before: only extrapolative during t = 0;
 rationality kicks in at t = 1 (“Minsky moment”
 Now: extrapolative also during bus

 Two state variables
 leverage (as before
 beliefs affected by past asset prices (new
 Two-dimensional nancial stabilit
 monetary policy alone: additional reason to lean
 against the wind at t = 0…
 …remains true with macropru policy
 fi
 )

 …

 y

 t

 )

 )

Monetary

 Monetary
 Macropr

 IT
 Rational
 IT +
 Expectation
 Macropru

 I Extrapolation
 Extrapolativ Lean Against during Bust
 Expectations Boom
 Macropru Lean Against
 Boom
+

+

 T

 u

 :

 e

Conclusion
 General theory of macropru + monetary polic
 workhorse for many applications
 general formula: MPCs and wedge

 Minksy Cycles with non-Rational Expectations
 expectation management: interventions attempt to
 mitigate nancial crashes in price
 dilemma: may affect monetary polic
 modi es optimal policy responses (targets and
 instruments)
fi
 fi

 s

 s

 y

 y