does assigning priority to deposits affect bank conduct? evidence from a quasi-experiment piotr...

34
Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Upload: maribel-sturmer

Post on 16-Dec-2015

219 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment

Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Page 2: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Introduction

Historically depositors have been protected during bankruptcy by having priority on claims (in the US national protection from 1993)

Following the Cypriot banking crisis the ECB has called for the introduction of depositor preference throughout the EU

Under Depositor Preference Legislation (uninsured) depositors have priority over general creditors on the residual assets of a failed bank

Proponents argue this will:

1. Prevent bank runs2. Engender more stable banking through increased market discipline

But the banking industry is sceptical arguing DPL will:

1. Raise costs for customers2. Destabilize the financial system 2

Page 3: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Introduction

We test the theoretical predictions made by Hardy (2013), Birchler (2000), and Osterberg (1996) regarding

Funding costs, liability structure, profitability, soundness, and valuation

Our research is important for three reasons

1. We inform an important policy debate that has far-reaching consequences that are not understood

2. Provide empirical tests relevant to theories on debt structure and monitoring (Fama, 1980; Goldberg and Hudgins, 2002; Rauh and Sufi, 2010; Hackbarth and Mauer, 2012)

3. Advance understanding of how bank capital structure responds to regulation (Le Lesle, 2012; Gropp and Heider, 2010)

3

Page 4: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

What do Depositor Preference Laws do?

The assets of a failed bank are paid out to creditors according to a claims structure

In most countries this looks like

DPL alters the claims structure by elevating uninsured depositors

4

Without DPL

1. Receiver2. Secured creditors3. Insured depositors4. Uninsured depositors &

general creditors5. Shareholders

With DPL

1. Receiver2. Secured creditors3. Insured & uninsured

depositors4. General creditors5. Shareholders

Page 5: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Theoretical Outline

Modigliani and Miller (1958) irrelevance theorem: absent taxation, the composition of corporate financing has no effects unless it:

1. Influences the probability of bankruptcy2. Affects the costs of bankruptcy

Hardy (2013) outlines a model of the bankruptcy process and the role of depositor preference within this

Creditors have a lobbying technology that is used to assert claims during bankruptcy proceedings

Lobbying is increasing in the residual assets of the failed bank5

Page 6: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Theoretical Outline

6

Because DPL assigns priority to some creditors, it reduces lobbying, and banks’ funding costs

Depositors no longer face losing their capital: demand a lower interest rate Reduces the expected costs of bankruptcy (through less lobbying)

The lower funding costs translate into higher profits (and increases firm value)

Endogenously lower the probability of bankruptcy

Because non-deposits are now junior, they demand higher interest rates (Osterberg, 1996)

Banks shuffle their liability structure

Birchler (2000) shows DPL leads to better/more efficient monitoring of bank conduct and risk taking

Page 7: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Theoretical Outline

7

Page 8: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Institutional Background

The assets of insolvent US banks are transferred to a receivership

The receiver’s task is to maximize the NPV of recoveries

The Banking Act of 1935 outlined a priority structure to the residual assets

1. Receiver2. Secured creditors3. Insured depositors {account balances < $100,000} (insured by FDIC)4. Uninsured depositors (account balances > $100,000) & non-depositors5. Holders of subordinated debt & shareholders

8

Page 9: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Institutional Background

30 US states opted to implement DPL between 1909 and 1993 (our analysis exploits 15 of these reforms)

9

Page 10: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Institutional Background

These state depositor preference laws elevated the priority of uninsured deposits

1. Receiver2. Secured creditors3. Insured and uninsured depositors4. Non-depositors5. Holders of subordinated debt & shareholders

Importantly (for our identification strategy) the state DPLs applied:

To state-chartered banks But not to nationally-chartered banks

10

Page 11: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Data Description and Representativeness

Quarterly Call Report data for commercial and savings banks in the US

Sample covers 1983Q1 to 1993Q2 for banks in 15 enacting states

11

Page 12: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Sample includes 199,698 observations for 5,506 banks

Broadly, the sample appears representative of the US population

12

Data Description and Representativeness

Page 13: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Measure funding costs as ratio of

Total interest expenses, deposit interest expenses, non-deposit interest expenses to total liabilities

Liability structure:

Total liabilities to total assets, same for deposits and non-deposits

Bank soundness:

Z-score, non-performing loans, leverage ratio

Profitability

ROA, total interest income to total loans, ROE 13

Data Description and Representativeness

Page 14: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Identification Strategy

We exploit plausibly exogenous variation in DPL enactment across states and time using a difference-in-difference estimator

We estimate

The dependent variable measures

Cost of funds; Liability structure; Bank soundness; Profitability

is a vector of bank-time varying controls; bank fixed effects; are state-quarter fixed effects

We therefore compare banks in the same macro environment14

Page 15: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Identification Strategy: Treatment Exogeneity

Why were the laws enacted?

Motivation for the reforms is not systematically documented

But like national DPL, adoption seems driven by the FDIC lobbying for DPL following the failure of Penn Square in 1982

DPL made bank resolution easier by allowing purchase and assumption transactions that minimize disruption of the local economy

Argued DPL would improve market discipline by exposing non-depositors to greater losses in the event of bankruptcy

Because of the limited discussion behind states’ enactment of DPL, we run a series of exogeneity tests 15

Page 16: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Identification Strategy: Treatment Exogeneity

Estimate a state-level Cox Proportional Hazards model of the form

Opt for a Cox PH model as it imposes no assumption on the shape of the hazard function, or the distribution of unobserved heterogeneity

Vector of controls includes: total failed banks’ deposits, estimated losses from bank failures, number of bank failures, HHI for bank deposits, total bank assets to GDP, and the unemployment rate

16

Page 17: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Identification Strategy: Treatment Exogeneity

No significant differences in any specification – DPL not influenced by our DVs – no simultaneity bias

17

Page 18: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Identification Strategy: Parallel Trends

To what extent do national-chartered banks act as a valid counterfactual?

18

Page 19: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Identification Strategy: Parallel Trends

Statistical tests confirm the graphical patterns

No significant differences in growth rates in the (immediate) pre-treatment period

19

Page 20: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Pricing Effects: Cost of Funds

Total interest expenses fall as deposits priority increases

Effect is equivalent to a 1.5% reduction in costs ($250,000 p.a.)

Heightened exposure to losses in bankruptcy causes an increase in non-deposit costs – evidence of increased market discipline 20

Page 21: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Quantity Effects: Liability Structure

Shrinking of state-chartered banks following treatment

Declines in deposits

But substitute towards non-deposit liabilities21

Page 22: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Quantity Effects: Liability Structure

What explains this behavior?

In equilibrium, and assuming risk neutrality, some depositors will move their deposits to CG banks that pay higher deposit interest rates

i.e. state-chartered banks lose out to nationally-chartered banks that were unaffected by DPL

TG banks make up for this by using more non-deposit funding

22

Page 23: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Quantity Effects: Market Shares

If our hypothesis is correct we should be able to document:

A decreasing reliance on deposit funding by state-chartered banks Particularly among uninsured deposits

23

Page 24: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Collateralization

Do non-depositors collateralize their claims?

Important as this could inhibit resolution costs and time No evidence for such phenomena

24

Page 25: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Soundness and Profitability

Policymakers emphasize that DPL will incentivize banks to operate safely, reducing the likelihood of failure

Non-depositors have stronger incentives to monitor banks’ risk exposure due to their junior claim

More skin in the game which makes them 1) withdraw funds; 2) refuse to roll over funds; 3) demand a higher risk premium; 4) demand collateral

These actions put constraints on the risk-taking behavior of banks’ asset allocation choices (Goldberg and Hudgins, 2002)

Market discipline consists of two dimensions

1. Monitoring: judge risk exposure and incorporate this information into security prices2. Influencing: claimants exert pressure on the bank to change conduct 25

Page 26: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Soundness and Profitability

Evidence suggests this view is correct

Improvement in soundness, reduction in NPL, and leverage

Increase in profitability, and a reduction in the variance of returns

26

Page 27: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Threats to Identification

27

Conduct a battery of robustness checks to ensure our results are not confounded by OVB

Because our regressions include state-quarter FE, these factors must be collinear with treatment to bias our findings

1. Charter is a choice variable A. Switching is infrequent (3.8%) B. No significant effect of treatment on charter type

2. Texas real estate collapse coupled with the Tax Reform Act of 1986 coincided with DPL in Texas (1985) Same findings when we omit Texas from the sample

Page 28: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Threats to Identification

28

3. Enactment of DPL in the northeast coincided with the New England banking crisis Results unchanged when we omit CT, ME, NH, and RI

4. If treatment is exogenous the magnitude of the ATE should be the same regardless of whether control variables are included or not This is indeed the case – very similar magnitudes

5. Placebo test – observed behavior should be specific to state-chartered banks and the actual treatment Randomly assign placebo treatments to national-chartered banks

at the time of DPL enactment: no significant effect

Page 29: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Threats to Identification

29

6. Bertrand, Duflo and Mullainathan (2004) Where the DV is serially correlated through time, DD estimators

will yield spuriously low standard errors Our regressions cluster at the “block” level (bank) Check robustness to myriad permutations of the error structure Same results when we cluster at the state level (only 15 groups!) We also collapse the data upon a pre- and post- treatment period

for each bank – similar results as before

Page 30: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Valuation Effects

A key prediction made by Hardy (2013) is that the reduction in funding costs raises firm valuations

This is consistent with the previous results

We now address how shareholders responded to DPL

This necessitates an alteration to the empirical methodology

Use event-study methodology to inspect what happened to stock returns following implementation of national DPL in 1993

30

Page 31: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Valuation Effects

National DPL was implemented following enactment of the Omnibus Reconciliation and Budget Act of 1993

The Act was devoted mainly to fiscal policy issues and balancing the government budget

National DPL went under the radar

It was also unexpected, even by regulators: the FDIC had to issue an emergency rule on how to interpret the legislation

Use stock price data retrieved from Datastream31

Page 32: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Valuation Effects

Run a time-series regression

Allow for AR(1) autocorrelation by applying Prais-Winsten adjustment prior to estimation

Focus on the event-day abnormal returns

32

Page 33: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Valuation Effects

33

Shareholders seem to welcome DPL

Positive valuation effects in line with the theory

Mostly similar evidence for recent EU and UK announcements

Page 34: Does Assigning Priority to Deposits Affect Bank Conduct? Evidence from a Quasi-Experiment Piotr Danisewicz, Danny McGowan, Enrico Onali and Klaus Schaeck

Concluding Remarks

The implications of changes in the regulatory environment on bank conduct are difficult to gauge

Our natural experiment yields important insights on this issue and theoretical models’ predictions

Robust causal evidence that DPL leads toLower funding costsImproved profitabilitySounder banksIncreased valuationNo change in collateralization

While the context may differ, the cleanness of our experiment helps shed light on an important EU-wide issue

34