Macro explanations for the crisis Popular views: the crisis was - - PDF document

6/11/2014 1

Cheap but Flighty: Global Imbalances and Bank Instability

Toni Ahnert

Bank of Canada

Enrico Perotti

University of Amsterdam

Macro explanations for the crisis

• Popular views: the crisis was caused by

– low interest rates (Fed policy), or – the accumulation of global imbalances.

• No clear foundations.
• The US credit boom was not funded by the

Fed, but by capital markets.

• Major role for capital inflows from EM

countries; may explain low rates

Capital flows between EM and rich countries

• Historically, funding moved to emerging

countries, though less than theory would imply (Lucas,1990). Why ? Political risk.

• From 1998, net flows from EM to rich

countries (especially US)

• Global imbalances = EM wealth rises

relative to the US

EM capital outflows

• EM savers target safe assets (Gourinchas Rey

2007, Caballero Krishnamurty 2009).

• Not just official reserves.
• Private flows appear to escape political and

currency risk

• How could this search for (anonymous) safety

cause instability ?

• We look at optimal bank funding in such a

framework

The Bright Side: Search for safety makes EM flows cheap

• Residents in EM seek safety in

countries with high protection of property rights

• Suppose their chance of expropriation

is p>0, so home storage yields 1-p

• Thus foreign savers offer cheap

funding

• In contrast, US savers can store safely

6/11/2014 2

Negative Rates of Return on Safe Assets

• Cattle common as investment in rural India
• Indian households earn negative average

returns of -64% on holdings of cows.

• If own labor priced at zero, average

return is -6%

• Indian explanation: we invest in them for

safety, banks are unreliable

• Negative returns as evidence of high

demand for safety

The dark side: Less informed

• Foreigners less able to assess local asset risk

(Rajan-Petersen; Stein; Cao Brennan)

• They may run even in the good state, causing

losses

• Trade off: cheaper but less stable funding
• Potential endogenous risk: when there is too

much unstable foreign funding, it leads also informed agents to run in the good state

Optimal funding choice

• Optimal contracting, no deposit insurance:

bankers pay for all risk

• But bankers have limited liability
• May trade off stability against funding

costs

• Result: private choice not socially optimal
• A key issue: what happens as foreign

wealth keeps increasing ?

Basic Model

• Domestic savers (mass 1-W) and foreign savers
• Intermediaries seek to invest in a LT project
• Return at time 2 is R with prob γ, else zero.
• At 1, future return known to banker & local

investors; foreigners less informed

• Asset may be liquidated at 1 (α <1)
• Optimal to do so when return is low

Information structure

• Three possible states at 1:

– G: Return R, confirmed by public signal (prob δ) – M: Return R, no public signal – L: Zero return, no public signal

• If no signal occur at 1, F expect state L with

increased probability (1-γ)/(1-γδ) > 1-γ.

• We rule out silly runs, when all savers know the

state is good

Social optimum in basic model

• The first best allocation is

– Full investment at 0 (I=1), attracting both domestic and foreign funding – Liquidation only in state L

• Intermediary should issue long term debt
• Drawbacks:
• Banker cannot discriminate foreigners to extract

safety rents

• LT debt claims are risky (since α <1)

6/11/2014 3

Main model: Safe Asset Bias

• Evidence of preferences for absolute safety

(Krishnamurty Vissing-Jorgensen 2013; Gorton, Lewellen, Metrick 2014)

• Evidence: a stable safe asset ratio in US
• When Treasury supply falls, private creation of

“safe assets” (deposits/AAA claims)

• A new banking model needed ?
• Agents seek a minimum value in all states

(Caballero Farhi, 2013; Gennaioli Shleifer Vishny, 2013)

• Contrast safety preferences with contingent

liquidity preferences (Diamond-Dybvig 1990)

Safe asset share in US wealth

Savers protect minimum wealth value

Extreme version: savers value only the safe amount (Gennaioli Shleifer Vishny 2013)

Seeking safety

• While all savers have storage, foreign savers may

be expropriated, domestic savers never. » Foreigners must satisfy their absolute safety demand S with an intermediary in a safe country.

• Let S = SF be the foreigners’ absolute safety

demand, while SD =0.

Short term debt for foreigners

• Foreigners want S with probability 1, so they

will not accept LT debt if α < S.

• New optimal funding:
• Bank offers both LT and demandable debt.
• Foreigners never take LT debt, domestic

savers take it if adequately rewarded

• In equilibrium, LT stable but more expensive

than ST debt.

Main result

• If foreign funding is cheap but not too

abundant, optimal contract is (expensive) long term debt for domestic savers and cheap demandable debt for foreign savers.

• Else the banker targets domestic funding by
• ffering only LT claims (mutual fund).
• This forces a smaller investment scale and a

higher funding cost.

6/11/2014 4

Notice that the required payment to long term debt D2 rises as w rises, as larger runs force to liquidate more in state M

LT to domestic, ST to foreigners

Payoff to informed savers

• Losses on foreign funding in M state, as it

runs too often

• Domestic savers insure foreigners,

demand higher return as R realized less

• ften
• Cost of informed funding increases in

foreign funding w

Payoffs What if w rises further

• At some point, it becomes impossible to

reward domestic savers for run losses

• Can bankers only target F ? (at cost of

smller scale)

• No! Because F does not invest in a bank

where only F invest !

• The banker must seek to secure enough

domestic savers (=stable funding) to attract the cheap foreign funding.

Upper bound on w

• If violated, return in the good state is not

enough to offer adequate compensation to domestic savers.

• But without LT savers, foreigners do not

invest either ! (since S>α)

Would bankers exclude cheap, unstable funding ?

• When excess liquidation costs in state G is

larger than NPV of more funding + extracted safety rent

• Not the same as social optimum
• Liquidation costs increasing in S and

decreasing in the precision of the signal δ.

6/11/2014 5

Self interested choice of bank capital

• Without deposit insurance, bankers

need to secure enough LT funding to cushion the bad state

• Voluntary capital ratios
• Paradoxically, partial deposit insurance

may reduce stability of funding

Conclusions

• Global imbalances imply more foreign wealth
• EM outflows into developed countries seek

safety, but are less able to asses local risk

• Cheap but unstable
• Shin’s proposal: tax non-core funding

Size of Global Imbalances

Text