Transcript Safety Nets, Prudential Standards, and Market Discipline

Banking Crises, Regulatory
Reform and Resolution
Charles Calomiris
May 11, 2010
Macroeconomics and Banking


View that banks are sources of shocks and
propagators of shocks in the macroeconomy was
“rediscovered” in 1980s, and has received more
attention in the past decade, especially after the
Asian and Mexican crises.
This view has substantial historical precedent,
and supporting historical evidence, and is only
“new” in the sense that it was ignored by most
macroeconomists in 1950s, 1960s and 1970s.
Why the Rediscovery?


Banking instability became more of an
issue in macroeconomics.
Key trends



Frequency of banking crises
Coinciding of banking crises/capital crunches
with macroeconomic declines
Mixing of banking crises with exchange rate
or sovereign crises.
EM Banks Especially Unstable



Fiscal costs of banking crises in EMs
amounted to about $1 trillion in the 1980s
and 1990s, which was equal to all foreign
assistance transfers from developed
countries from 1950-2001.
Many of these collapses also involve twin
crises (collapses of both exchange rate
and banking systems), with dire
macroeconomic consequences.
Why is this happening? What can be done?
Central Questions about Bank Risk and
Prudential Regulation





Why are crises so common and so severe
now?
Are banking crises and their causes the same
or different from those of the past?
Are crises inherent in the function and
structure of banks?
Which supervision and regulation rules work?
Should government pursue counter-cyclical
forbearance policies, or procyclical prudential
regulation?
Bank Function and Structure


Banks control risk via intensive screening,
contracting, and monitoring, involving private
information production.
Bank function and structure






Delegated monitoring on asset side, liquidity creation
on liability side
Maturity and liquidity transformation
First-come-first-served rule
Essential structure and function of banks has
created a special role for debt market discipline
of banks, in contrast to other firms.
Does this make banks risky? Does that explain
EM banking crises?
Alternative view: Insider lending, moral hazard,
politics of bailouts (Populism meets cronyism)
Crises have different shapes

Currency depreciation only, reflecting an overvaluation of
the currency, but without effects on banking system


Banking collapse only


Brazil 1999: overvalued currency, lack of fiscal discipline, but banks
relatively healthy, no resurrection risk betting
Australia 1893: costs of losses in banks absorbed by stockholders
and depositors, not taxpayers
Twin crises, in which banking collapses and exchange rate
collapses happen together. The new phenomenon of twin
crises (only a handful were observed in pre-1980 era). But
twin crises can also be distinguished according to the
dominant direction of causation


fiscal crisis => banking instability as in Argentina 2001, where
government lacked the safety valve of inflationary monetary policy,
and thus stole bank capital as last resort means of financing;
or banking instability => fiscal crisis, as in Mexico 1994 and East
Asia 1997, although in these cases there was also an overvaluation
problem of real exchange rate, in Mexico because of boom in
demand, and in Asia because of cumulative decline in productivity
Financial system feedback during crises
Currency
devaluation
Rapid rise in
interest rates
Capital outflow
Overextended
banks
and firms
Risk factors show themselves during crises





Law: inability to have orderly workout in financial distress
leads to backlog of unresolved debts; reversal of
privatization contracts, redenomination of contracts, limits
on capital flows, nationalization of assets.
Information: markets react dramatically to crises (shut
down of orderly information processing, high adverse
selection costs)
Fiscal policy: lack of political will to reduce expenditures,
improve tax collection, avoid inflation tax.
Banking: desire to protect bankers, who are often borrowers
and political allies, too, leads to lack of credible discipline ex
ante, and big bailouts ex post. Quasi privatization can be
worse than public banks from the standpoint of the severity
of crises.
KEY POINT: All these risks can be observed in advance of
the crisis!

Recent crises forecasted in particular countries
Mexico: The dog that did not bark in December 1994.






East Asia: Diminishing returns / cronyism





Weak bank balance sheets, high corporate leverage were known
Low or negative return on invested capital due to crony banking
Overvaluation related to declining productivity
Recessions reflecting declining productivity, overvaluation
Brazil: Insufficient fiscal reform



Fiscal spending and bank lending, election year
Overvaluation (Dornbusch)
Bank privatizations, lack of recapitalizations
Recession began in 1993
Sterilization, reserve outflows, tesobonos liquidity risk
Overvaluation
Old story of unsustainable peg given rising inflation
Argentina: Insufficient fiscal reform



Coparticipation
Labor, tax policies, recession (overvaluation-induced deflation)
Destruction of banking sector
Causation in Mexico







Spending, weak privatized banks (Haber article), offbalance sheet exposures raise debt and expected
monetization.
As reserves are drained, central bank sterilizers,
thereby increasing current money supply.
Rise in money and expected money drive up prices,
leading to overvaluation and increasing pressure on
exchange rate.
Zedillo’s non-reform leads market to realize
inevitability of collapse, and run begins.
FX exposure is combination of direct bets and
defaults linked to dollar debt.
Banks double bets on FX by having lots of both (in
violation of regs, done with swaps via Wall Street).
Uncanny (forecasted) replay of Chile 1982-83 crisis.
Rising Debt, Sterilization
Sterilization had been the rule
Inflation and base growth linked
Inflation increasingly threatened exchange rate
Dec-96
Oct-96
Aug-96
Jun-96
Apr-96
Feb-96
Dec-95
Oct-95
Aug-95
Jun-95
Apr-95
Feb-95
Dec-94
Oct-94
Aug-94
Jun-94
Mexican Pesos / U.S. Dollar
Mexican Devaluation
9
8
7
6
5
4
3
2
1
0
Asian Crises




March and April 1997, the Economist and FT
have special reports on declining fortunes of
Asian banks, and possible crises there.
Alwyn Young writes about declining
productivity as threat to sustainability of socalled Asian miracle in 1994-95.
Short-term borrowing in dollars increases as
risk rises (largely interbank, “protected”?,
and with different weights according to
Basel).
IMF assistance bails out those debts.
Bank Losses in Asian Crises
Thailand Indonesia Korea
1997NPL/TL
19%
17%
16%
1997NPL/GDP
30%
10%
22%
Cleanup Cost /
1999 GDP
42%
55%
20%
Diminishing Returns => Increasingly Inefficient
Capital Investment in East Asia in 1980s, 1990s
Return on Capital
Employed Minus
Interest Rate, 1992
Indonesia
Korea
Malaysia
Philippines
Thailand
-12%
- 3%
3%
-13%
- 9%
Source: Pomerleano (1998)
Micro Level Stylized Facts
As the 1990s progressed . . .
•Asian corporations experienced a decline in
performance.
•Asian corporate managers increased the
leverage of their firms.
•Asian corporate managers borrowed
substantially from international capital markets
in foreign currencies (US Dollars).
How Can You Bet the Country?





Government bailouts are anticipated, intermediated by
government’s relationship with the IMF, which injects
dollars to government, which pays them to crony firms
with outstanding short-term debts.
Taxpayers pick up the pieces.
High leverage of ex ante insolvent banks and firms
indicates that both borrowers and US, Japanese, and
European bank lenders anticipated this.
Note: Capital flows, per se, are not the problem, but
rather the allocation of risk by government associated
with those flows. There is an argument for waiting to
liberalize capital flows until incentives and financial
regulation have been fixed.
This is a particularly important issue for China, given
that it could repeat the Asian crisis pattern, given
diminishing returns and lack of market discipline in
banking system.
Trends in Corporate Leverage Ratios
Country Comparisons
Median Leverage Across Countries
250
200
1992
1996
150
100
50
T ha
iland
an
Taiw
p in e
s
Phil
ip
aysia
Mal
Kor
ea
I ndo
nesia
0
Rating
Ratio
AAA
13.4
AA
21.9
A
32.7
BBB
43.4
BB
53.9
B
65.9
Brazil’s Controlled Devaluation (Healthy banks)
Argentina’s Crisis Anticipated
Interest Rates on 30-Day Time Deposits in Pesos and Dollars
60
50
Interest Rate (%)
40
Pesos
Dollars
30
20
10
Jan-02
Dec-01
Nov-01
Oct-01
Sep-01
Aug-01
Jul-01
Jun-01
May-01
Apr-01
Mar-01
Feb-01
Jan-01
Dec-00
Nov-00
Oct-00
Sep-00
Aug-00
Jul-00
Jun-00
May-00
Apr-00
Mar-00
Feb-00
Jan-00
0
Sources: J.P. Morgan Chase & Co.; Banco Central de la República Argentina, Interest Rates on Deposits (available at http://www.bcra.gov.ar/).
Argentina’s Crisis Anticipated
Difference Between Interest Rates on 30-Day Time Deposits in Argentina in
Pesos and Dollars vs. EMBI+ Argentina Strip Spread
30
5000
20
4000
Weekly Interest Rate Difference
Daily EMBI+ Argentina Strip Spread
15
3000
10
2000
5
EMBI+ Argentina Strip Spread
Difference in Interest Rates (%)
25
6000
1000
0
Jan-02
Nov-01
Sep-01
Jul-01
May-01
Mar-01
Jan-01
Nov-00
Sep-00
Jul-00
May-00
Mar-00
0
Jan-00
-5
Sources: J.P. Morgan Chase & Co.; Banco Central de la República Argentina, Interest Rates on Deposits (available at http://www.bcra.gov.ar/).
Deposit Outflows
Monthly Dollar Deposits in Argentina, 2001
54
53
52
Billion U.S. Dollars
51
50
49
48
47
46
45
44
Dec-01
Nov-01
Oct-01
Sep-01
Aug-01
Jul-01
Jun-01
May-01
Apr-01
Mar-01
Feb-01
Jan-01
43
Source: Argentina Ministry of Economy & Production, Macroeconomic Statistics (available at http://www.mecon.gov.ar/peconomica/basehome/infoeco_ing.html).
International Reserve Outflows, 2001
40
35
Billion Pesos
30
25
20
15
10
5
Dec-01
Nov-01
Oct-01
Sep-01
Aug-01
Jul-01
Jun-01
May-01
Apr-01
Mar-01
Feb-01
Jan-01
0
Note: Because of a change in the BCRA’s definition of international reserves, data after October 31, 2001 includes public bonds involved in reverse
repo-operations. Data before October 31 does not include these bonds.
Source: Banco Central de la República Argentina, International Reserves and BCRA’s Financial Liabilities (available at http://www.bcra.gov.ar/).
Annual Capital Inflows by Type
40
Billion U.S. Dollars
30
Total
FDI
Equity
Debt
Other
20
10
0
-10
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
-20
Note: Equity inflow was not available until 1992. I use the following variables from the IMF’s International Financial Statistics as components
of capital inflows:
1) FDI: line 78bed (Direct Investment in the Reporting Economy, n.i.e.)
2) Equity: line 78bmd (Equity Securities Liabilities)
3) Debt: line 78bnd (Debt Securities Liabilities).
4) Other: line 78bid (Other Investment Liabilities, n.i.e.)
Source: International Monetary Fund, International Financial Statistics, June 2004.
Dec-02
Oct-02
Aug-02
Jun-02
Apr-02
Feb-02
Dec-01
Oct-01
Aug-01
Jun-01
Apr-01
Feb-01
Dec-00
Oct-00
Aug-00
Jun-00
Argentine Pesos / U.S. Dollar
Argentine Devaluation
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Role of fixed exchange rates



All of the problems discussed would still
be problems under flexible exchange rates
But fixed exchange rates make things
worse by create sudden adjustments, and
thus big accumulations of risk.
This not only causes sudden problems, it
also worsens resurrection risk taking by
giving banks and firms something to bet
on.
Panics vs. Insolvencies


Concern that too much, unwarranted, sudden
market discipline can create undesirable social
costs from contraction of bank deposits during
“panics” is the primary justification for bank
safety nets in theory and in history (deposit
insurance, and lender of last resort).
Such systemic panics (as distinct from periods of
high bank failure) resulted from a combination
of observable shocks and unobservable
incidence of shocks, in combination with the
structure of banks (liquidity transformation, fcfs
rule).
U.S. Experience with Panics



1857, 1873, 1884, 1890, 1893, 1896, 1907
Observable shock was a dual threshold of 9%
stock market decline over a quarter, and 50%
increase in seasonally adjusted liabilities of failed
business (NEWS RELEVANT FOR BANKS)
Some shocks originated in NYC and were related
to securities markets, use of funds by NYC banks:
In 1857, loans to bond dealers, connections to
RRs, was the problem. Some shocks may relate
more to peripheral areas (perhaps 1893).
Dealing with Panics


Costly bank panics were almost exclusively
a U.S. phenomenon by the mid-19th century.
Market discipline, along with inter-bank
cooperation and lending, central banks, and
clearing house actions to share risks dealt
with threat of panics effectively, except in
U.S. where branching limits, pyramiding of
reserves created concentrations of risk, and
made coordination difficult.
Desire to keep unit banking, and risk of
panics explains why U.S. originates deposit
insurance.
Dealing with Panics (Cont’d)




Banks assisted each other, sometimes
through formal clearing house actions.
Market discipline kept risky banks in check,
and made other banks see advantage to
identifying and punishing risky banks
quickly.
Suspension was used as a last resort, and
market discipline created incentives to
restore convertibility quickly.
Resumption of convertibility would occur
when secondary market discounts on bank
paper approached zero.
Historical Banking System Collapses


Panics in U.S. did not produce banking
collapses, because the combination of
market discipline, clearing house support,
and temporary suspension of convertibility
(until asymmetric information was
resolved) insulated banks from costs.
1893, worst of U.S. panics, coincided with
large exogenous agricultural problems, but
bank failures produced negative net worth
of failed banks of only 0.1% of GDP.
Historical Collapses (Cont’d)


Worldwide, from 1873 to 1913, there were
no more than seven episodes of severe
bank failure worldwide (defined as
collapses that produced banking losses
where negative net worth of failed banks
in a country exceeded 1% of GDP)
Argentina 1890 (~10%), Australia 1893
(~10%), Norway 1900 (~3%), Italy 1893
(~1%), Brazil various (hard to measure,
but all much less than 10%). Only 2-3 of
these are “twin crises.”
Historical Comparison between Today
and Pre-WWI Era Appropriate?


1870s-1913 is a time when capital flows to
emerging markets were high relative to GDP,
limited liability banking was growing rapidly
around the world, countries relied on fixed
exchange rates, and macroeconomic climate
was very volatile.
This suggests that according to some
explanations of crises (exchange rate fixed,
free chartering of banks, multiple equilibria
due to foreign capital flows) we should see
more then than now. But we do not.h
Historical Collapses (Cont’d)



Land booms and busts underlay these collapses, and
often bank risk was subsidized by government in one
way or another. Argentina, Italy subsidized risky bank
lending on land; Norway and Australia promoted land
booms in other ways.
Banking collapses for some U.S. states also directly
traced to safety net policies. Agriculture boom and
busts and banking collapses were much more severe
in states with deposit insurance (WWI price bets).
Twin crises in Italy and Argentina in 1890s reflected
feedback from banking crises to fiscal collapse of
government (foreshadowing today’s crises).
State-Level Deposit Insurance in 1920s
3 Insured 15 Controls
Asset Size
Equity / Assets
Growth during Boom
Loans / Assets
Negative NW of fails
/ Survivors NW
Source: Calomiris JEH 1990.
$320
0.11
185%
0.76
$622
0.13
128%
0.70
3.5
0.5
How the Safety Net Causes Bank Collapses




Safety net removes market discipline that used
to operate, both as a check on conscious risk
taking, and on quality of bank management
making risk taking decisions. Both effects are
important.
These two channels do not operate with
constant adverse effects, but rather, their
effects vary over the cycle.
Conscious risk taking increases in wake of losses
(resurrection bets on unlikely outcomes with
high risk premia, especially in currency markets,
which deepens extent of twin crises through
feedback effects).
Management quality problem can be most
hazardous during booms, and becomes visible
during busts (WWI grain price bets).
Why Few Twin Crises Historically?




Other than those exceptions, fiscal discipline
coincided with and reinforced benefits of
market discipline over banks.
Governments adhering to gold standard had
access to international capital markets, and
could act to protect banks with classical
lender of last resort liquidity assistance.
Mexico 1907 and Russia 1899-1900 are
prime examples of successful assistance
Assistance was limited by credible
commitment to stay on gold standard, which
in turn ensured access to funds as needed.
(Contrast to IMF)
What About the Great Depression?



New research (Calomiris and Mason, 1997,
2003) shows that panics were not nationwide
phenomenon until very late (early 1933)
For the most part, fundamental shocks
(deflationary monetary policy, gold standard,
agric. distress, other bad economic policies)
caused insolvencies by many banks, not panics.
And, despite the severe shocks and many
failures, losses of failed banks 1930-1933 only
about 3-4% of GDP.
Banking Collapses Today



In contrast, about 150 episodes since 1978 of banking
system collapses with costs of more than 1% of GDP,
more than 20 with costs in excess of 10% of GDP, and
many of those have costs in excess of 20% of GDP.
This is unprecedented. Collapses often coincide with
currency collapse due to fiscal implications of banking
collapse for government. This reflects changes in
political economy of banking systems (similar to
Eichengreen 1996 argument on inflation process).
Like Italy and Argentina pre-WWI, these severe
collapses have been directly traced to
incentives from government policies protecting
banks from market discipline.
How Did / Do Disciplined Systems Behave?

Old-Fashioned Disciplined Banking



Equity/Assets and Asset risk managed to target
low default risk on debt of bank. During good
times, equity capital is cheap (no lemons
problems) and lending opportunities are good,
so both risk and equity capital rise.
When shock hits, banks face prospect of loss of
deposits due to combination of risk aversion and
need for liquidity of depositors, and asymmetric
information problem about losses within bank.
As banks lose deposits they act to restore
confidence by contracting loans, cutting
dividends, and expanding cash asset holdings.
NYC Bank Capital and Risk 1920-1936
NYC Banks’ Loans/Cash, Equity, Dividends
Loans/Cash
1922
2.1
Equity/Assets
0.18
1929
3.3
0.33
1933
1.0
0.15
1940
0.3
0.10
Source: Calomiris-Wilson JB 2004.
Dividends
$392m
$162m
Discipline Reflected on Liability Side

If discipline exists, it appears in three forms:





Interest cost of debt goes up with risk
Rationing effect: deposits decline
Shift to high-cost, “monitored” marginal funds
These effects are consistently visible
historically, as well as currently, in all types
of countries.
Bank liability data, and liability interest rate
data are the most reliable, consistently
reported data on balance sheets, which helps
make them especially useful as indicators.
Example: Chicago 1932
Number
1932 Failures
46
1931 RD
1932 Survivors
62
2%
1%
1931 Borr/Debts
12%
2%
1931 Dep growth
-45%
-33%
Source: Calomiris-Mason 1997 AER.
Example: Argentina 1995
1995 Failures 1995 Survivors
RD paid in 1993
13%
9.5%
Example: Mexico 1996


Even though there was 100% deposit
insurance, the losses were so large, and
the political debate so uncertain, that
insured deposits were not necessarily
protected.
Banamex (marginally solvent) paid 17%
on its funds, on average, but Bank Serfin
(deeply insolvent) paid 29%
Market Discipline Even with Insurance
All banks Banamex Serfin
1996 Dep. Int.
25.2%
17.4%
28.9%
1994 branches
1996 branches
5,051
6,264
710
912
561
578
Contrast Old (Stable) System with
Protected (Unstable) System


Old, disciplined system reduced bank risk
in response to shocks.
New system sees banks increasing risks in
response to shock (doubling their bets),
especially taking on exchange risk.





Chile 1982-1983
U.S. Savings and Loans in 1980s
Mexico 1993-1994
Japan 1990-1997
Korea, Thailand, etc. 1995-1997
Risk-Based Capital Regulation


Basel and similar systems supposedly target
risk-based capital, which is similar to
targeting default risk of debt, although
measurement of risk and capital are
imperfect (to say the least) and ratios are
arbitrary.
If this is successful, it results in regulatory
discipline with effects similar to market
discipline: When a bank loses capital, it
contracts risk, cuts dividends, in order to
comply with standard and in doing so
reduces default risk.
Procyclical Effects?


If effective, this sort of capital regulation
necessarily exacerbates the business cycle,
since losses of capital (during onset of
recession) produce contraction of bank loan
supply, which aggravates the recession.
Calomiris and Mason (AER 2003) estimate
that market discipline during U.S. Great
Depression was responsible for income
declines roughly one-third as large in
percentage terms as the percentage declines
in loan supply. These are very large effects.
Procyclical Effects (Cont’d)



Two points warrant emphasis:
First, any prudent risk-managing bank will have
to contract risk in response to loss, so
procyclicality is an inevitable feature of a wellmanaged banking system.
Second, “forebearance” (the decision by
regulators to relax requirements because of a
concern about loan supply) tends to produce a
larger procyclical effect, because
resurrection risk taking leads to systemic
collapse, and often unproductive risk taking
prior to collapse.
Policy in the Real World





Even though policy makers are aware that forbearance is
counterproductive, they still do it.
Politics tends to produce strong incentives for protection of
banks and forbearance, more in some countries than in
others (Demirguc-Kunt, Kane and Laeven 2007).
Book capital requirements invite discretionary forbearance,
as do reliance on supervisory judgments when measuring
risk and capital.
Also, policymakers may lack timely information, or incentive
to put forth effort to collect it, so some forbearance may be
inadvertent.
S&R fails. Barth-Caprio-Levine find that regulatory and
supervisory practices, other than practices that
introduce market discipline, make no difference for
banking sector growth or stability. Market discipline
promotes both greater stability and higher growth.
The Public
Corruption
Media
Politicians
corruption
Regulators and
supervisors
corruption
Banks
The Market:
Depositors,creditors,
rating agencies
Borrowers,
counterparties
Technology, Information
Infrastructure
A Framework for Bank Regulation,
Barth Caprio and Levine
Market Structure
Judicial, Legal,
Regulatory Environment
Institutional Environment
Democratic, Political Structure/System
Barth et al Book



Study of regulatory practices across 150
countries, asking whether regulations improve
bank stability and growth; and also exploring
how regulations reflect and alter political
economy and corruption.
Variation in regulatory practice is enormous
(capital regulation, deposit insurance coverage,
government ownership of banks, foreign entry
permitted, bank powers allowed).
Evidence favors the “private interest” (or
“grabbing hand”) view over the “public interest”
(or “helping hand”) view of bank regulation and
supervision in EMs.
Barth et al Findings
“Across the different statistical approaches, we
find that empowering direct official supervision of
banks and strengthening capital standards do not
boost bank development, improve bank efficiency,
reduce corruption in lending, or lower banking
system fragility. Indeed, the evidence suggests that
fortifying official supervisory oversight and
disciplinary powers actually impedes the efficient
operation of banks, increases corruption in lending,
and therefore hurts the effectiveness of capital
allocation without any corresponding improvement
in bank stability.
In contrast to these findings…bank supervisory and
regulatory policies that facilitate private sector
monitoring of banks improve bank operations.”
Barth et al Details: Univariate Regressions
Dependent Var: Bank Credit to Private Sector/GDP
(Controls not reported)
Entry Req. Index
Limits on For. Entry
Entry Appl. Denied
Activities Restrictions
Capital Regulations
Prompt Corrective Power
Official Supervisory Power
Supervisory Independence
Government-Owned Banking
Private Monitoring Index
Negative Insignificant
Negative Significant
Negative Significant
Negative Significant
Positive Marginal
Negative Insignificant
Negative Significant
Negative Insignificant
Negative Significant
Positive Significant
Barth et al: Multivariate Regression
Dependent Var: Bank Credit to Private Sector/GDP
(Controls not reported)
Entry Req. Index
Activities Restrictions
Capital Regulations
Official Supervisory Power
Government-Owned Banking
Private Monitoring Index
Positive Insignificant
Negative Significant
Positive Insignificant
Positive Insignificant
Negative Insignificant
Positive Significant
Barth et al: Multivariate Regression
Dependent Var: Bank Crisis Probability (logit)
(Controls not reported)
Entry Req. Index
Activities Restrictions
Capital Regulations
Official Supervisory Power
Government-Owned Banking
Moral Hazard Index
Political Openness
MH x PO
Positive Insignificant
Positive Significant
Negative Insignificant
Negative Insignificant
Positive Marginal
Positive Significant
Positive Insignificant
Negative Significant
Moral Hazard = Dep Insurance Generosity
Barth et al: Multivariate Regression
Dependent Var: Corruption of Banking Officials
(Controls not reported)
Official Supervisory Power
Positive Significant
Government-Owned Banking Positive Significant
Private Monitoring
Negative Significant
Other Evidence on Foreign Bank Entry




Greater supply of credit (Goldberg, Dages, Kinney
2000), although based more on “hard” information
than “soft” information (Mian 2003)
Less local presence => greater volatility of credit
(Herrero and Peria 2005)
Giannetti and Ongena (2005) find that foreign
presence reduces connected lending problems,
improves access of funds to efficient nonconnected borrowers, and improves efficiency of
capital allocation, although effect seems confined
to medium and large firms
Similarly, Bonin and Imai (2005) show that sale of
Korean banks to foreign lenders had large
negative effects on stock returns of related
borrowers.
% Banking System Foreign-Controlled (IMF 2000)
Czech Rep.
Hungary
Poland
Argentina
Brazil
Chile
Colombia
Mexico
Peru
Venezuela
Korea
Malaysia
Thailand
1994
5.8
19.8
2.1
17.9
8.4
16.3
6.2
1.0
6.7
0.3
0.8
6.8
0.5
1999
49.3
56.6
52.8
48.6
16.8
53.6
17.8
18.8
33.4
41.9
4.3
11.5
5.6
How can bank regulatory policy help?



Repeal the safety net. Perhaps a good idea (a
vast body of empirical evidence indicates deposit
insurance makes systems more prone to failure,
not less), but this is hard to do, and there is often
implicit insurance even when explicit insurance is
repealed.
Narrow banking. This is repeal in disguise.
Internal Models. Regulators’ incentives are still to
forbear, and thus credibility is lacking. Regulation
and supervision become extremely complicated. For
developing countries, the cost of implementing
complex regulatory apparatus is very high, because
banking systems are small (100 countries with
banking system deposits under $10 billion).
Combining Safety Nets, S&R, and Elements
of Market Discipline


Once the government is involved in protecting
banks, markets have little incentive to take risks,
and thus little incentive to monitor, or to act upon
information.
S&R can incorporate market discipline if it finds a
way to (1) require banks to offer some component
of risky debt to finance themselves, (2) ensure that
the pricing of that instrument is observable to them
and the market, and (3) establish rules that force
S&R to act upon the information produced by the
market.
What Is NOT Market Discipline



Disclosure rules, by themselves, are of little use,
since market may have little reason to care about
disclosure.
The presence of uninsured debt is not necessarily
indicative of market discipline because the debt
may effectively have the option to leave when
things go wrong, or because government may use
mergers or other protection to bail it out, justified
by “least cost resolution policy”
Insiders’ holdings of debt (which can be hard to
track), or outsiders’ holdings protected via
derivatives written by the bank, won’t provide
discipline.
Getting Market Discipline to Work: The
Integrated Approach of Argentina c. 1999


An integrated approach to S&R, in which credible
market discipline is required, and produced by a
combination of reforms, is the best approach.
One of the best examples of this approach was
Argentina in the late 1990s. The collapse of the
system in 2001-2002 was not due to regulatory
failure, but rather its success. Cavallo seized bank
equity in 2001 to alleviate the government’s fiscal
problems because domestic banks were only place
to find the money.
Argentina 1992-2000







Free foreign entry (competitive pressure, skills)
Encouragement of privatization of loss-making
provincial banks (pay provinces to privatize and
renounce bank chartering). 18 privatized 1992-99.
No explicit deposit insurance (modified in 1995)
Book equity capital requirement depends on loan
interest rate
VAR to set capital requirement for market risk
based on market volatility
Liquidity requirement can be satisfied with standbys
(rewards banks that command market confidence)
Aggressive NPL policy
Argentina (Cont’d)





BASIC System: Integration of rules for information
creation, disclosure, and use.
Central de riesgo information pool (crafted to
minimize free riding, while allowing banks to avoid
bad credits, and avoid double pledging of
collateral).
Auditing supervised, bonded.
Subordinated debt requirement forces banks to
issue 2% of deposits in the form of uninsured
subordinated debt held at arms length.
(Concentration of uninsured claim may be
desirable, as more informed, and harder to renege
if amount is small…no systemic excuse.)
Banks rated by approved rating agencies (good
intentions, bad outcome).

Ratings reform proposals today.
Argentina (Cont’d)


Subordinated debt is not occurring in a vacuum,
but rather alongside other capital and liquidity
rules, and other rules, that give market opinions
power in the regulatory process.
Problems with implementation of sub debt rule.




Lack of compliance penalized in theory, but not clear
whether penalized in practice.
No clear regulatory actions (e.g., closure) required
based on high yields or inadequate issuance. No public
information on yields, amount, compliance.
Not exempted from least-cost resolution.
Arms length holding not enforced effectively.
Argentina (Cont’d)


Still, there is evidence that sub debt helped
Low compliance was indicative of bank weakness,
and central bank realized this, so it gave them a
signal, but not one that the public had access to
(so no discipline on regulatory forbearance).
Source: Calomiris-Powell 2001.
RD
NPL
Equity ratio
1996-99
High compliance
7%
14%
0.157
1996-99
Low compliance
8%
25%
0.183
Argentina (Cont’d)



More generally, Argentina showed healthy
signs of operating with prudent (oldfashioned) risk management.
One indication of the effectiveness of the
system is the fact that deposit growth rates
reflect deposit risk, and that deposit risk is
related to book measures of asset risk and
equity capital.
Another indication of effectiveness is that
banks that experienced increases in their
interest cost of debt acted quickly to reduce
risk, and thus bring interest cost back down.
Argentina (Cont’d)
Dependent Variable: Quarterly Deposit Growth
Regressor
Coefficient
Stand.Error
Eq Ratio (-1)
Loan Int. Rate
Loans/Cash
0.277
-0.254
-0.0032
Sample period: 1993:3-1999:1
Number of Observations: 1,138
Adjusted R-Squared: 0.31
0.074
0.121
0.0007
Argentina (Cont’d)
Deposit Interest Rate Autoregression
Dependent Variable: Quarterly Deposit
Interest Cost
Regressor
RD (-1)
Coefficient
Stand.Error
-1.29
0.04
Adjusted R-Squared: 0.58
Number of Observations: 688
Broad Conclusions




Safety net is major source of risk, market
discipline is the ONLY credible mitigator of
risk.
Market discipline is best introduced with an
integrated approach that results in the
creation, disclosure and use of credible
information in the marketplace.
Government regulatory transparency is as
important as government mandates for
promoting market discipline.
Market discipline is a complement, not a
substitute for S&R (e.g., insider holdings).
Specific Recommendations

Good ideas for bringing markets into S&R process:
 Free entry, privatization of SOBs, limited safety net
 Use of loan interest rate to set capital requirement
 Reform of credit ratings: numbers with penalties.
 Standbys abroad should substitute for cash reserves
 BASIC Argentine system (except use of credit ratings)
 Improve on Argentine sub debt to avoid back door
bailouts, insider holdings, derivatives; to enforce public
disclosure; and to establish prompt corrective action rules
to limit forbearance in response to failure to gain market
confidence (defined by market yields, flows on debt)
 Sub debt’s form can be flexible (e.g., two year maturity
CDs held by banks abroad) depending on environment,
and contingent capital certificates are an improvement on
sub debt (discussed further below).
Cutting Edge Ideas
CoCos







Require on top of higher common/assets minimum
requirement of 12%, a CoCo requirement of 10% of
quasi market value of firm (face value debt +MVE)
Trigger must be credible, predictable, timely, and
based on comprehensive view of bank.
Goldman proposes book value, but it is not credible or
timely.
Cumulative market value decline trigger would work
(say, 40% from peak).
Dilution risk would force voluntary preemptive issues
of common stock ahead of triggers.
Result would be that banks almost never go under.
Design Prompt Corrective Action (PCA) trigger based
on similar cumulative value decline basis.
Market Cap for Large American Financial
Institutions
120
100
Index, Mar-07 = 100
AIG
Bank of America
80
Bear Stearns
Citigroup
60
Lehman Brothers
Merrill Lynch
40
Benchmark
Contingent Capital Trigger
Wind-Down Trigger
20
0
Mar-07
Jun-07
Sep-07
Dec-07
Apr-08
Jul-08
Oct-08
140
Market Cap for Large European Financial
Institutions
120
Index, Mar-07 = 100
100
Dexia
80
Fortis
ING
60
Lloyds
RBS
40
UBS
20
0
Mar-07
Jun-07
Sep-07
Dec-07
Apr-08
Jul-08
Oct-08
Macro Prudential Rules


Macro prudential regulation that raises capital
requirements during normal times in order to
lower them during recessions.
Additional macro prudential regulatory
triggers that increase regulatory
requirements for capital, liquidity, or
provisioning as a function of credit growth,
asset price growth, and possibly other
macroeconomic risk measures. (Borio
Drehman paper.)
Case Study: Colombia 2006-2008







Financial system loans annual growth rose from 10% in December
2005 to 27% by December 2006.
Core CPI rose gradually relative to credit (from 3.5% in April 2006 to
4.8% in April 2007).
Real GDP growth in 2007 was 8%.
Current account deficit rose from 1.8% GDP in second half of 2006
to 3.6% GDP in first half of 2007.
Monetary authority reacted directly to credit growth in real time:
Interest rates were increased 400 bps from April 2006 to July 2008.
But central bank saw too small a market response to this, so it
 increased reserve requirements for banks and
 convinced superintendency to raise provisioning for credit,
 imposed measures to raise costs of borrowing short-term from
abroad (deposit requirement reactivated), and
 limited not only currency mismatches of banks and other FX
exposure in the system, but also gross currency positions (to
avoid counterparty risks).
Credit growth is now “only” 13%; risk-weighted capital ratio for
banks is 13.9%, and first half 2008 is 4.9% above first half of 2007,
expected to fall to about 3.5% for 2008 as a whole.
Fix Too Big To Fail




CoCos for large banks probably will take care of much of
the problem.
Regulatory surcharge (which takes the form of higher
required capital, higher required liquidity, or more
aggressive provisioning) on large, complex banks.
Detailed regularly updated plans for intervention and
resolution of large, complex institutions prepared by
them, which specify how control the bank’s operations
when transferred to a prepackaged bridge bank if the
bank became severely undercapitalized.
Hybrid reliance on bankruptcy with special
resolution authority triggered by credible
determination of real systemic risk. Key problem:
how to keep unwarranted resolutions from happening?
Aftermath of crises

Does crisis tend to lay groundwork for more or
less liberalization?


It can go either way, depending on domestic political
environment (Mexico, East Asia and Brazil vs.
Argentina)
How to deal with massive insolvencies?

There are many possible mechanisms







Liquidation via market (but adverse selection problems, legal
system and information limits)
AMCs a solution? No, since they also must liquidate, and
same problems plague them; also corruption plagues them,
much more than in U.S. or Scandinavia.
Creative solutions that work with market incentives (Punto
Final program in Mexico)
Assistance to recapitalize banks (RFC vs. Japan)
Foreign entry of banks to speed reestablishment of credit.
Debt moratorium (cultura de no pago in Mexico)
Debt redenomination (Argentina; U.S. precedents)
US Bank Assistance in 1930s vs. Japan’s in
1990s (w/ Mason)

US Banking problems




Real shocks, not runs (until 1933) (CalomirisMason 2003a)
Market discipline  Capital crunch (CalomirisWilson 2004)
 Depositor preferences (active vs. passive)
 Loan supply contracts
 Dividends cut
Large adverse macroeconomic consequences of
credit contraction (Calomiris-Mason 2003b)
Asset market illiquidity problems, too, from
liquidation of bank assets, which slowed resolution
US in 1930s (Cont’d)

Policy Response


RFC lending (inadequate)
RFC preferred stock begins in 1933
 Selective: Targeting marginal banks, field
office autonomy seems to have limited abuse
 Limits behavior: Dividends, capital, voting on
management issues; Regression evidence
suggests that RFC conditionality mattered
 Seems to be effective, reduces failure risk
Comparison
RFC Recipients Non-Recipients
Fail Prob ’31
Fail Prob ’34
0.147
0.011
0.096
0.004
Div Pay ’34
0.001
0.008
Japan in the 1990s





Little deposit market discipline
Convoy system spreads assistance
Dividend payments to common stock
remain large (liquidity for Keiretsu firms)
Banks do not have to accumulate
additional capital to get assistance
Assistance seems to have had no effect on
failure risk or lending, and NPLs have
grown over time
What might have worked better?

Conditional Assistance



Dividend limits
Capital plan with matching requirement (note
that this is self-selecting)
Introduce Market Discipline in Regulation



Minimum sub debt or other uninsured debts
Interest rate-sensitive capital requirements
High reserve requirements, but allow offshore
SLOCs in lieu of reserve requirements
Debt redenomination: US in 1860s, US in 1930s




In both cases, this was a convenient way of
using the change in numeraire of debt to
restore net worth of banks and borrowers.
Civil War: legal tender acts saved banks after
shock of December 1861 by making bank
liabilities decline with bank assets (govt. bonds).
Great Depression: elimination of gold clauses
actually increased values of bonds, implying
that effect on default premium outweighed
effect on face value.
Key advantages: speed, no reliance on
institutional quality.
Argentina’s Redenomination 2002




Does same logic apply to Argentina 2002?
Fiscal crisis leads to arm twisting of banks to
buy bonds, then eventual collapse of banks
and exchange rate.
“Pesification” of debt could provide relief to
borrowers, especially high dollar debt firms in
non-tradables sector.
Asymmetric aspect of pesification hurt banks
and had clear political element, but we can
conceptually separate it from pesification.
Argentina (Cont’d)





Argentina’s peculiar vulnerability to
devaluation
Investment evidence
Stock market reactions
Conclusion: There seems to have been a
significant positive effect on market and on
non-tradable producers with high dollar debt
exposures.
Net benefit?
Figure 1C: Reinhart-Rogoff-Savastano Composite Dollarization Index Level vs.
Exports-to-GDP Ratio for Reinhart-Rogoff-Savastano’s
Fifty Highest Dollarized Countries
1.0
Estonia
0.9
Bahrain
0.8
Angola
2001 Exports/GDP
0.7
Belarus
Thailand
Mongolia
Tajikistan
Hungary
0.6
Bulgaria
Vietnam
0.5
Cambodia
Ghana
Moldova
Philippines
Croatia
Turkmenistan
Jordan
Indonesia
Costa Rica
Jamaica
Guinea-Bissau
Honduras
Kyrgyz RepublicYemen
São Tomé & Príncipe
Côte d'Ivoire
Russia
Turkey
0.4
0.3
El Salvador
Armenia
Georgia
0.2
Pakistan
Uganda
0.1
Guinea
Malawi
Congo DR
Sierra Leone
Tanzania
Mozambique
Zambia
Paraguay
Ecuador
Uruguay
Bolivia
Lebanon
Peru
Argentina
0.0
8
10
12
14
16
18
20
22
Reinhart-Rogoff-Savastano Composite Index Level: 1996-2001
24
26
Non-Tradable Firms
Mexico
Tradable Firms
Mean
Firms
Std. Dev
Mean
Firms
Std. Dev
0.0711
7
0.0669
0.1460
23
0.1460
0.1625
28
0.2141
0.1224
25
0.0893
HighDollar
Debt
Firms
LowDollar
Debt
Firms
Dif
-0.0914
0.0236
T-Stat
-1.9155
0.6695
Pr(Dif ≥ 0)
0.0324
0.7463
Non-Tradable Firms
Tradable Firms
Mean
Firms
Std. Dev
Mean
Firms
Std. Dev
HighDollar
Debt
Firms
LowDollar
Debt
Firms
0.0724
6
0.0959
0.1817
7
0.1579
0.0644
5
0.0397
0.2155
7
0.1537
Dif
0.0080
-0.0339
T-Stat
0.1862
-0.4069
Pr(Dif ≥0)
0.5712
0.3456
Argentina
DDAt-1
DDAt-1 x ARG
TR x DDAt-1
TR x DDAt-1 x ARG
-0.1449
-1.79
(0.0809)
0.076
0.2192
1.63
(0.1346)
0.106
0.1242
1.21
(0.1029)
0.230
-0.2684
-1.64
(0.1638)
0.104
EVENT STUDY
Dependent Variable
Cumulative Raw
Returns
Independent Variables
TR
DDAt-1
Constant
Cumulative Abnormal
Returns
Coefficient
(Std. Error)
T-Stat
P>|t|
Coefficient
(Std. Error)
T-Stat
P>|t|
0.1561
3.82
0.1249
2.37
(0.0409)
0.001
(0.0528)
0.026
0.2447
2.33
0.3473
2.56
(0.1051)
0.028
(0.1356)
0.017
-0.0255
-0.49
-0.0959
-1.42
(0.0524)
0.631
(0.0676)
0.168
Regression Statistics
Observations
29
29
R2
0.4070
0.2912
Adjusted R2
0.3613
0.2367
Root MSE
0.1009
0.1301