Why Don - PowerPoint PPT Presentation

1 / 14
About This Presentation
Title:

Why Don

Description:

Introduction Prior Research Findings Criticisms of Prior Research Question Under Study Sample and Data Sample and Data Logistic Regression Multi-State Model Model ... – PowerPoint PPT presentation

Number of Views:23
Avg rating:3.0/5.0
Slides: 15
Provided by: Michael3086
Category:

less

Transcript and Presenter's Notes

Title: Why Don


1
Why Dont All Troubled Banks Fail?
  • Oshinsky, R. and Olin, V. (2006).
  • Troubled Banks Why Don't They All Fail?,
  • FDIC Banking Review (Vol. 18, pp. 22).
  • Michael Campbell
  • Public Policy Analysis (ECON 539)
  • 02/02/09

2
Introduction
  • Many failure-prediction models and early-warning
    systems already exist.
  • Focus has been on pairs of outcomes
  • Failure vs. non-failure
  • Merger vs. consolidation
  • Benefits to FDIC and bank researchers.

3
Prior Research Findings
  • Specific findings include failure results from
    low capital, embedded risky assets, poor
    management, low earnings, and low liquidity.
  • Most findings are binary, i.e. this or that, e.g.
    failure v. non-failure, acquisition v.
    consolidation, etc.

4
Criticisms of Prior Research
  • Rarely in life are possible outcomes binary.
  • Benefits are diminished by not studying other
    possible outcomes.
  • Data set includes all banks vs. only troubled
    banks.
  • Limited prediction accuracy and inherent bias.

5
Question Under Study
  • Troubled Banks
  • Why Dont They All Fail?

6
Sample and Data
  • Data sources CAMALS ratings from regulatory
    agency and quarterly FDIC CALL reports.
  • 1,996 banks on the FDIC problem-bank list from
    1990 2002.
  • Event pairing provided 3,747 observations.
  • Observation events Event 1 - CAMELS rating of a
    4 or 5, and Event 2 bank recovery, merger or
    consolidation, problems remain, or failure.

7
Sample and Data
  • Dependent Variables
  • Succeed, Fail, Remain Troubled, and
    Merge/Consolidate
  • Independent Variables
  • Capital Ratio, Asset Quality, Management
    Efficiency, Earnings Ratio, and Liquidity Ratio.

8
Logistic RegressionMulti-State Model
  • Univariate Trend Analysis Used to determine
    whether prior-period financial characteristics
    differ by future bank state.
  • One-way Analysis of Variance Used to examine the
    financial characteristics of recovered banks
    versus banks in the other three states.
  • Multinomial Logistic Estimating Procedure Used
    to model future bank states. The model
    simultaneously estimates three binary logits for
    comparison among the outcome categories.
  • The general form of the model tested
  • Probability of State (k)i,t F(Financial
    conditioni,t-1, Economic conditionst)
  • (Probability of State (k)i,t is the probability
    that bank i will be in state k at time t)

9
Model Explained
  • Multinomial Logistic Model is a regression model
    which generalizes logit regression by allowing
    more than two discreet outcomes.
  • Used when the dependent variable(s) is nominal
    (succeed, fail, remain unchanged,
    merge/consolidate) and when the response is not
    ordinal (cant be ordered 1, 2, 3).

10
Model Assumptions
  • Examples
  • Each independent variable has a single value.
  • The dependent variable(s) cannot be perfectly
    predicted from the independent variables.
  • Collinearity is relatively low (improved
    predictability of a variables impact upon y).

11
Models Effectiveness
  • Models effectiveness was evaluated by
  • Comparing results to competing models.
  • Investigating the economic and statistical effect
    of the explanatory variables.
  • Verifying that banks with the highest
    predictability of failure actually did fail.

12
Results
  • General agreement with researchers expectations
  • Improved net income ratios associated with
    recovery state.
  • Reduced non-performing assets and expenses
    associated with recovery state.
  • Reduced exposure to volatile liabilities and
    assets associated with recovery state.
  • No expectations with financial ratios for merged
    or remain troubled states except that theyll
    fall between succeeded and failed bank ratios.
  • Positive coefficient signs success Negative
    coefficient signs failure.

13
Conclusions
  • By deviating from previous models this study
    focused on only troubled banks resulting in
    greater predictability.
  • This four-state model offers the FDIC and bank
    researchers more information
  • Eliminates bias by excluding healthy banks.
  • Identifies alternative outcomes.
  • Enables better estimate of contingent loss
    reserve.
  • Better long-term strategic resource planning.
  • The additional information can better assist the
    FDIC is long-term strategic planning.

14
Policy Implications
  • Because the model shows that certain explanatory
    variables affect future bank states this can
    assist regulators in choosing policies that
    affect the likelihood that troubled banks can
    successfully resolve their own problems.
Write a Comment
User Comments (0)
About PowerShow.com