the role of political frictions in financial crises · the role of political frictions in financial...
TRANSCRIPT
Politics & Economics of International FinanceMarch 28th, 2015
The Role of Political Frictions in Financial Crises
Francesco TrebbiUniversity of British Columbia,
CIFAR, NBER
Build-up to the Crisis
Aftermath of the Crisis
The Role of Political Frictions in Financial Crises
2
The Role of Political Frictions in Financial Crises
Build-up to the Crisis
Aftermath of the Crisis
Politics After the
Crisis
3
Build-up to the Crisis
Aftermath of the Crisis
Politics After the
Crisis
2 Specific Instances of Response to Crisis
Dodd-Frank Act of 2010
EU Banking Union of 2012
The Role of Political Frictions in Financial Crises
4
Build-up to the Crisis
Aftermath of the Crisis
Politics After the
Crisis
Dodd-Frank Act of 2010
EU Banking Union of 2012
The Role of Political Frictions in Financial Crises
McCarty, Poole, Rosenthal Political Bubbles 2013Rajan Fault Lines 2010
Johnson, Kwak 13 Bankers 2011
Igan, Mishra, Tressel NBER Macro Annual 2012
Igan, Mishra JLE 2015
Posner, Weyl AER P&P 2013
Frieden JIMF 2015
5
Build-up to the Crisis
Aftermath of the Crisis
Politics After the
Crisis
Bertrand, Bombardini, Trebbi 2015
Dodd-Frank Act of 2010
EU Banking Union of 2012
The Role of Political Frictions in Financial Crises
Agarwal, Lucca, Seru, Trebbi QJE 2014
Mian, Sufi, Trebbi QJPS 2013 Mian, Sufi, Trebbi AER 2010
Mian, Sufi, Trebbi AEJ Macro 2014
6
Focus on aftermath of financial crises (banking, currency, debt, inflation).
Known stylized facts: Deep economic contractions (output and employment). Reinhart
and Rogoff (2009), Reinhart and Reinhart (2010).
Sustained waves of volatility, often resulting in ‘secondary’ crises (e.g. debt crises following banking crashes Reinhart and Rogoff, AER
2011).
This is not all. Economies polarize: 1997 Asian Crisis: Korea experienced a 5% increase in Gini from
1996 to 1998 (Cheong, 2005). 4% in the Philippines. Klein and Shabbir (2006, ch.1).
World Bank (2000) reports increases in inequality in 15 out of 20 crisis episodes in Latin America.
Politics After the Crisis
7
Mian, Sufi, Trebbi (2014) show that not only economic, but political polarization systematically increases around financial crises:
Voters become more ideologically polarized; Government coalitions become weaker: in terms of both vote shares & seat shares;
Oppositions become larger; Party fragmentation increases across the board.
Political gridlock & lack of reform/intervention is a natural outcome.
(i.e. the failure of the US Congressional Supercommittee on deficit is the norm, not the exception).
Politics After the Crisis
8
Example: Post-Crisis Decreases Mass at Ideological Center
.05
.1.1
5.2
Avg.
Sha
re in
Ideo
logi
cal B
in
0 2 4 6 8 10Ideology (Right is 10)
Pre-Crisis Post-Crisis
Banking Crisis
Notes: All 70 Reinhart and Rogoff (2011) Countries. All Crises 1975-2010. Pre-Crisis Sample: 5 years before first year of crisis. Post-Crisis Sample: 5 years after last year of crisis.Self Positioning in Political Scale, World Values Survey 1981-2008 Official Aggregate (e033, 2009).
9
Example: Post-Crisis Decreases Mass at Ideological Center
.05
.1.1
5.2
Avg.
Sha
re in
Ideo
logi
cal B
in
0 2 4 6 8 10Ideology (Right is 10)
Pre-Crisis Post-Crisis
Banking Crisis
Notes: All 70 Reinhart and Rogoff (2011) Countries. All Crises 1975-2010. Pre-Crisis Sample: 5 years before first year of crisis. Post-Crisis Sample: 5 years after last year of crisis.Self Positioning in Political Scale, World Values Survey 1981-2008 Official Aggregate (e033, 2009).
10
1. Crises bring gridlock through polarization. Any model of post-crisis macro intervention that leaves this political feature aside forgoes an important dimension. Any type of reform becomes harder. Including bailouts.
2. Crises are occasionally thought of as critical junctures where macroeconomic reform unlocks by shattering entrenched conditions. Benefits of crises: Drazen & Grilli (AER 1993); Drazen & Easterly (E&P 2001). The opposite seems true.
3. Gridlock delays reform & possibly makes recovery slower (explains long recessions).
4. Gridlock brings political uncertainty. Markets for sovereign debt do not like that → Debt crises. E.g. EU.
5. Gridlock brings selective intervention. If a reform overcomes political gridlock, it’s because of strong political organization. Organized special interests (Banks) get a sizeable bailout. Diffused special interests (Mortgage debtors) don’t. Olson (1965).
Why is this important?
11
Application & Methodology Pre- & Post-Crisis samples. 5 years windows around Reinhart and
Rogoff’s (2011) crises.
Within-country analysis (country F.E.) & time F.E. (accounts for cross-border contagion, possibly too harsh a constraint).
Employ survey data on individual ideological positioning. (E.g. World Values Survey panels; ANES; etc). Data coverage is not good unfortunately.
Employ electoral and political data from Database of Political Institutions (World Bank) & IMF Reform Database. Data coverage is excellent.
12
0.01
.02.0
3.04
Den
sity
20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Banking Crisis
0.01.
02.03.
04.05
Den
sity
0 20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Currency Crisis
0.005.0
1.015.0
2.025
Den
sity
0 20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Domestic/External Debt Crisis
0.0
1.02
.03
Den
sity
0 20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Inflation Crisis
Notes: All 70 Reinhart and Rogoff (2011) Countries. All Crises 1975-2010. Pre-Crisis Sample: 5 years before first year of crisis. Post-Crisis Sample: 5 years after last year of crisis.Opposition share from Database of Political Institutions (World Bank, 2010).
Post-Crisis Decrease in Majority Margins for Government.
13
0.01
.02.0
3.04
Den
sity
20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Banking Crisis
0.01.
02.03.
04.05
Den
sity
0 20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Currency Crisis
0.005.0
1.015.0
2.025
Den
sity
0 20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Domestic/External Debt Crisis
0.0
1.02
.03
Den
sity
0 20 40 60 80 100Vote Share of Government Parties
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Inflation Crisis
Notes: All 70 Reinhart and Rogoff (2011) Countries. All Crises 1975-2010. Pre-Crisis Sample: 5 years before first year of crisis. Post-Crisis Sample: 5 years after last year of crisis.Opposition share from Database of Political Institutions (World Bank, 2010).
Post-Crisis Decrease in Majority Margins for Government.
14
Table 6
Banking Crisis Currency Crisis Dom./External Debt Crisis Inflation Crisis
(1) (2) (3) (4) (5) (6) (7) (8)Dependent Variable: Government Vote Share
Post-Crisis
-10.6029 -6.8459 -5.6889 -2.9830 -17.0451 -3.3900 -26.6331 -10.2615
[1.4469]** [1.4906]** [1.4648]** [1.0052]** [2.8974]** [2.3458] [2.9077]** [1.6419]**
R2 0.09 0.67 0.03 0.77 0.13 0.84 0.27 0.92
N 534 534 599 599 236 236 279 279Dependent Variable: Opposition Vote Share (Excluding Unaligned Parties)
Post-Crisis
8.6544 7.7531 2.8580 0.5635 10.9867 2.5713 20.4801 6.3344
[1.5059]** [1.3673]** [1.5110] [1.0068] [2.7145]** [2.6374] [2.8892]** [2.1033]**
R2 0.06 0.71 0.01 0.75 0.07 0.74 0.17 0.86
N 534 534 599 599 236 236 279 279Dependent Variable: Polarization
Post-Crisis
0.1761 0.1002 0.0971 0.0605 0.2732 0.1126 0.4836 0.1099
[0.0625]** [0.0637] [0.0646] [0.0489] [0.0753]** [0.0840] [0.0616]** [0.0727]
R2 0.01 0.64 0.00 0.63 0.03 0.57 0.09 0.67
N 752 752 753 753 366 366 411 411
Note: Columns (2), (4), (6), and (8) include country and year fixed effects. Robust standard errors in brackets. ** Significant at .01 * significant at .05. Includes only country and year observations within 5 years before or after a crisis.
Weak Governments After Crises
15
01
23
Den
sity
0 .2 .4 .6 .8 1Fractionalization Index
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Banking Crisis
01
23
Den
sity
0 .2 .4 .6 .8 1Fractionalization Index
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Currency Crisis
0.5
11.5
2D
ensi
ty
0 .5 1Fractionalization Index
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Domestic/External Debt Crisis
01
23
4D
ensi
ty
-.2 0 .2 .4 .6 .8Fractionalization Index
Pre-CrisisPost-Crisis
Note: Kolmogorov-Smirnov p-valueof Rejection of Equality of Distributions = .001
Inflation Crisis
Notes: All 70 Reinhart and Rogoff (2011) Countries. All Crises 1975-2010.Pre-Crisis Sample: 5 years before first year of crisis. Post-Crisis Sample: 5 years after last year of crisis.Fractionalization Index from Database of Political Institutions (World Bank, 2010).
Post-Crisis Increase in Party Fractionalization in Legislative
16
Reform After the Crisis: It’s there but it’s weak & rare We perform same event study methodology, but looking at
reforms.
Focus on any change in i) the degree of liberalization of interest rate controls; ii) directed credit/reserve requirements; iii) entry barriers/pro-competition measures; iv) privatizations; v) capital account restrictions; vi) banking supervision; and vii) security markets liberalizations. See Abiad et al., 2008
Large reforms (either liberalizations or retrenchments) appear rare. Roughly 1 in 10 crises produces a sizeable response (e.g. case of creditor rights reform).
17
Poster Child of Post-Crisis Reform: Dodd-Frank Act of 2010
“Passing this bill was no easy task. To get there, we had to overcome the furious lobbying of an array of powerful interest groups and a partisan minority determined to block change.”
President Barak Obama at Dodd-Frank signing ceremony July 21, 2010
Source: whitehouse.gov
18
Poster Child of Post-Crisis Reform: Dodd-Frank Act of 2010 Largest regulatory intervention in finance & banking since 1930s. Statute is 848 pages. Includes 398 Rulemaking Requirements. As of 2015,
Finalized Rules run around19,000 pages.
Covers: i. Creating Financial Stability Oversight Councilii. Regulatory reorganization (OTS dissolved) iii. Securitization Reform (“Skin in the game for mortgage originators”)iv. Derivatives Regulation (CFTC oversight, Clearing Requirements)v. Creating Consumer Financial Protection Bureauvi. Rating Agency reformvii. Limits to Proprietary Trading (“Volcker Rule”)viii. Executive Compensationix. Capital Requirements (esp. Banks > $50 Billion in assets)
For entities above $50 Billion in consolidated assets Dodd-Frank imposes “enhanced minimum standards for compliance programs”
Bertrand, Bombardini, Trebbi 201519
Bank Capitalization: Tier1 of Top-4 Banks
68
1012
1416
2000q1 2005q1 2010q1 2015q1Tier 1 Risk Ratio: Core Capital (Tier 1)/Risk-Adjusted Assets
NBER Recession ALL BANKS, MEANJPMORGAN CITIGROUPBANK OF AMERICA WELLS FARGO
20
Bank Capitalization: Tier1 of Banks >$50B
-50
5C
ontra
sts
of L
inea
r Pre
dict
ion
2000q1 2005q1 2010q1 2015q1Tier 1 Risk Ratio: Core Capital (Tier 1)/Risk-Adjusted Assets
Differential behavior of Banks >$50 Bil
21
Charge-offs by Top-4 Banks: Who Performed the Worse in the Crisis?
0.0
05.0
1.0
15
2000q1 2005q1 2010q1 2015q1Qr. Net Charge Offs as a percentage of average loans
NBER Recession ALL BANKS, MEANJPMORGAN CITIGROUPBANK OF AMERICA WELLS FARGO
22
Charge-offs by Banks >$50B: Who Performed the Worse in the Crisis?
-.002
0.0
02.0
04C
ontra
sts
of L
inea
r Pre
dict
ion
2000q1 2005q1 2010q1 2015q1Qr. Net Charge Offs as a percentage of average loans
Differential behavior of Banks >$50 Bil
23
Regulatory Rulemaking Process
Proposed Rule
Interim Rule
Finalized Rule
Proposed Volcker Rule
298 pages11/7/2011
LAW
Finalized Volcker Rule1077 pages12/10/2013
Volcker Rule
11 pages7/21/2010
24
Dodd-Frank Act of 2010: Rulemaking Completion
020
4060
8010
0
% F
inal
ized
Rul
es O
ver T
otal
, Dav
is-P
olk
2010m1 2011m1 2012m1 2013m1 2014m1 2015m1
Share of Finalized Rules of Dodd-FrankRulemaking Requirements
Dodd-Frank Act Signed
25
Measuring Banks’ Influence on Regulators
Proposed Rule
Interim Rule
Finalized Rule
Meetings & Comments17,000+ Comments on Volcker Rule
(250 by banks)
26
Dodd-Frank Act of 2010: Role of the Banks
According to disclosed commentary and meeting information roughly 2% of total comments come from banks.
However, about 90% of meetings of regulators are with bankers/bank lobbyists.
Things we do:
Assess special interests role in regulation:
E.g. Compute the amount of regulatory text in Finalized Rule that cannot be traced back to corresponding Interim or Proposed Rules.
Which are the banks that comment on the rules that end up changing the most?e.g. Goldman Sachs’ influence is determined by whether systematically Goldman Sachs’ comments on rules that change a lot.
27
Dodd-Frank Act of 2010: Bank Lobbying Expenditures
020
0040
0060
0080
00
2000q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1
Annual Lobbying Expenditures (in $1000) for banks >$10 Bil
NBER Recession ALL BANKS, MEANJPMORGAN CITIGROUPBANK OF AMERICA WELLS FARGO
28
Correlation between Lobbying & Charge-offs in the Crisis
Note: log(total amount lobbied between 2012 and 2015). Losses are proxied by total net charge offs between 2008-2012 as share of avg. assets held between 2008-2012. Correlation conditional on: i) log(avg. assets held between 2008-2012); ii) log(total amount lobbied between 2000 and 2007).
-10
-50
510
e( ln
_am
ount
_lob
_pos
t | X
)
-.05 0 .05 .1e( losses | X )
coef = 39.604557, (robust) se = 17.074275, t = 2.32
29
Correlation between Regulatory Influence & Charge-offs in the Crisis
-10
010
20e(
influ
ence
| X
)
-.05 0 .05 .1 .15e( losses | X )
coef = -35.893982, (robust) se = 14.814294, t = -2.42
Note: Influence is proxied by the number of mentions of comments from the bank in interim or final rulemaking documents. Losses are proxied by total net charge offs between 2008-2012 as share of avg. assets held between 2008-2012. Correlation conditional on: i) log(avg. assets held between 2008-2012); ii) log(total amount lobbied between 2000 and 2007).
30
Correlation between Regulatory influence & Tier1 in the Crisis
-10
010
2030
40e(
influ
ence
| X
)
-10 -5 0 5e( tier1 | X )
coef = .95739389, (robust) se = .2832868, t = 3.38
Note: Influence is proxied by the number of mentions of comments from the bank in interim or final rulemaking documents. Tier1 is the avg. tier1 capital ratio between 2008-2012. Correlation conditional on: i) log(avg. assets held between 2008-2012); ii) log(total amount lobbied between 2000 and 2007).
31
● Possible silver lining of the crisis: Birth of EU-wide financial regulation in large scale.
● European Banking Authority (EBA), created in 2010, in charge of bank stress tests.
● European System Of Financial Supervision (ESFS). Coordinating financial services supervision (banking, securities, insurance) across Eurozone/EU.
But Banking Prudential Supervision rested with National Authorities
→Not Integrated within EU27 nor Euro17→Not Fit for Cross-border Entities
Reform After the Crisis: EU Banking Union of 2012
32
Sept 2012, European Commission: Proposal for a Single Supervisory Mechanism (SSM) for banks.
New Role for ECB as Super-National Prudential Regulator:→Licensing/Authorizing→Assessing qualifying holdings→Ensuring compliance in regulatory capital requirements→Carrying out preemptive intervention measures.
But:→On-site examinations left to National Supervisors (w/ ECB “opt-in”)→National Supervisors still assess validity of internal risk models (for assets risk weighting for regulatory ratios).
Much of what follows based on: Agarwal, Lucca, Seru, Trebbi (QJE 2014) & Lucca, Seru, Trebbi (JME 2014).
SSM
33
The US As Laboratory for the EU: Overlapping US Financial Regulators
34
Example: US State-Chartered Banks
35
CAMELS Upgrades/Downgrades
CAMELS upgrade CAMELS downgrade
Freq. Percent Freq. PercentFederal
Regulator 1332 45% 3665 62%State
Regulator 1619 55% 2281 38%
Total 2951 100% 5946 100%
Mean SD Mean SD∆CAMELS -1 0 1.13 0.38
36
CAMELS upgrade CAMELS downgrade
Freq. Percent Freq. PercentFederal
Regulator 1332 45% 3665 62%State
Regulator 1619 55% 2281 38%
Total 2951 100% 5946 100%
Mean SD Mean SD∆CAMELS -1 0 1.13 0.38
CAMELS Upgrades/Downgrades
Federal regulator twice as likely to downgradethan State
37
CAMELS upgrade CAMELS downgrade
Freq. Percent Freq. PercentFederal
Regulator 1332 45% 3665 62%State
Regulator 1619 55% 2281 38%
Total 2951 100% 5946 100%
Mean SD Mean SD∆CAMELS -1 0 1.13 0.38
CAMELS Upgrades/Downgrades
Somewhat countered by upgradesby State
38
Heterogeneity Across States: Regulatory “Spreads”
-.1
0
.1
.2
.3
Coef
ficien
t, st
ate-
FEDS
inte
racti
on
AL AR AZ CA CO CT DE FL GA IA ID IL IN KS KY LA MA
MD
ME MI
MN
MO MS MT
NC ND NE NJ NM NY OH OK OR PA SC SD TN TX UT VA WA WI
WV
WY
State
39
Additional Findings
Banks respond to differential regulatory behavior. Federal regulators induce readjustments of
→Tier1 capital ratios, → Leverage, → NPLs & Delinquencies, → Implying lower ROA.
State-Fed regulatory “spreads” vary across states. Larger spreads correlate/predict
→ Higher frequency of bank failures,→ More Problem banks,→ Slower TARP repayment,→ Costlier resolutions.
40
Why do these differences exist? Explaining Federal/State differences:
Local regulators protect local constituents Higher spread during “tougher” times Higher spread for privately funded banks
Regulatory capture Limited support that higher spread in states with higher corruption Limited support for “revolving door”
Competence/Funding of resources: Higher spread in states with lower movement into private sector Higher spread in states with lower training budget Higher spread in states with lower # of examiners per manager
41
Politics in the aftermath of financial crises systematically different. i.e. US Debt Ceiling-type gridlocks not the exception.
Case #1: Dodd-Frank. Large, sweeping reform post crisis. Has it been systematically gamed by
special interests?
Case #2: SSM. Has EU rushed into a potentially inferior regulatory architecture in the
aftermath of a sovereign debt crisis?
Main take away: Political/politico-economic frictions play massive role in aftermath financial crises.
Cannot be disregarded in the economics of macro response to crisis. But, with rare exceptions, they are.
Conclusion
42