Financial Innovation, Macroeconomic Stability and Systemic Crises Discussion of

advertisement
Discussion of
Financial Innovation,
Macroeconomic Stability and
Systemic Crises
Arvind Krishnamurthy
Northwestern University

Questions:




Effects of financial innovation on macroeconomic
stability.
Systemic risk/crises.
Costs and benefits of innovation
Relevant questions

Geithner quote about credit derivatives
Outline



Paper writes down a stylized model to help
think about these questions
My discussion
Explain the mechanics that drive the model



Externality/ policy application
Non-monotonic effect of maximum loan to value/
leverage ratio ( ) on probability of crises
How does the model shed light on the effect
of innovations, such as credit derivatives
Liquidity at date 1
In order to avoid liquidation at date 1,
intermediaries/firms need to raise at least
( - xs + b1s ) i0
xs < 0 is cash need for business operation
b1s > 0 is debt repayment
 Raise money in two ways:



Sell capital ( )
Borrow against capital ( )
Selling Capital: Fire Sales
Raise ( - xs + b1s ) i0 by selling ks units at price q
ks = ( - xs + b1s ) i0 / q
q
kD ()
Fire Sales and Financial Depth ()

Less financial depth: Demand curve not perfectly
elastic

Multiple equilibria / liquidation externality
Policy: Role for the Fed LLR to rule out bad
equilibrium?
Same downward sloping supply idea has been used
to discuss Bank Runs, 1987 market crash, 1997
Asian Crisis, 1998 LTCM episode …


Fire Sales and Financial Innovation ( )



Constraint on raising debt:
(at Date 0)
b1s <  q1s
(at Date 1)
b2s <  q2s
Raising  loosens date 1 constraint, ks down
Raising  also allows for more borrowing at
date 0: b1s up and i0 up
Cash need is ( - xs + b1s ) i0
Non-monotonicity
Date 0

Agents don’t internalize liquidity externality in
date 0 decisions


Over-leveraging: too much i0 / too much b1s
Policy: Restrict ex-ante leverage
Financial Innovation: Raising 


Raises probability of crises over part of the
range
Improvements in automobile safety:




Air bags
Crumple zones
Anti-lock brakes
People drive faster …


Perhaps more accidents
But we may on net be better off
Is Raising  beneficial for Welfare?


Something the paper should compute
Unclear, for two reasons



On average beneficial (borrowing at date 1)
Paper shows externality, but does not show
whether externality increases with 
My guess: In model, the Date 1 effect
dominates because it helps across all states
of the world, whereas date 0 effects are only
in the crisis/liquidation states.
Raising : Crisis costs vs. average benefits


Conjecture: I think the model could say that
an LLR can solve the date 1 externality, so
that LLR plus  increase is beneficial
Possibly unambiguous in this case, which
would give a different angle on the Fed’s role.
Model and Reality

Financial innovation in the model is about
changing 


Problem model identifies is that agents overleverage, not internalizing crisis costs


Reality: options, credit derivatives, MBS, etc.,
Reality: Risk management of financial firms.
``Insufficient” risk management.
Let us think about the risk management of a
new financial asset, i.e. credit derivatives
Financial Innovation is about the New


Financial innovations are complex even to
sophisticated market participants
Risk management of an unknown product




Learning
Model risk
Interpreting and acting on outcomes that the
model does not expect
Examples



1987 Stock Market Crash
1997 Asian Crises
1998 LTCM Crisis
1987 Stock Market Crash


Insurance Strategies: Synthetic Puts/
Portfolio Insurance
Pre-1987: Compare implied volatilities on outof-the-money put options to at-the money
options (Bates 2000)


Post-1987: Same comparison


Out-of-the moneys have 3% higher implied vols
Out-of-the moneys have 10% higher implied vols
The crisis led agents to better understand the
cost of crises
1997 Asian Crises


Integration into world capital markets: Foreign
borrowing, capital inflows
Pre-1997: Liability composition tilted towards




Post-1997:




Dollar-denominated debt
Short-term debt
Portfolio inflows
Reserve accumulation
Long-term debt
FDI flows
Post-crises, countries adopted insurance strategies
1998 Hedge Fund Crisis




Hedge funds become the central liquidity
providers in many specialized asset markets
(MBS, sovereign debt, equity vol, …)
Pre-crisis risk management: Stress testing
based on historical correlations
Crisis: Liquidity shortages cause unusual
comovement.
Post-crisis risk management: Stress testing
scenarios include liquidity events
Credit Derivatives

Geithner (2006) quote that begins this paper



“…uncertainty about how exposures will evolve
and markets will function in less favorable
circumstances”
Example: GM and Ford downgrades from last
year
What are the true exposures and how will
markets play out?
Average Benefits versus Crisis Costs
Model and My Take on Reality






In the model, innovation is an increase in allowable
leverage ()
Financial innovation is about the new
Realized events do not span the entire event space
Uncertainty about outcomes
What is the best crisis risk management strategy?
Conjecture: Systemic risk is always in the new
assets, and not in the old assets (i.e. portfolio
insurance is well understood at this point).
Average Benefits versus Crisis Costs
Model and My Take on Reality




Date 1 policy in model: Eliminate liquidation
equilibrium (role for an LLR)
I agree, but with a slightly different angle:
The liquidation externalities arise when bad
outcomes are realized, and agents are not
prepared
Perhaps also some “over-reaction” by agents

As in this paper: Role for an LLR
Average Benefits versus Crisis Costs
Model and My Take on Reality


Date 0 policy in model: Reduce overleveraging
More generally, this translates to: Improve
date 0 risk management


What is date 0?
Unclear if anyone (including the Fed) knows the
best strategy
Conclusion

The relevant calculation is average benefits
versus crisis costs
There are liquidation externalities

Innovation is about the new…

Download