STATISTICAL TECHNIQUES – Project topics materials

Get Complete Project Material File(s) Now! »

Lau (1987) and Ward (1994)

Both Lau’s (1987) and Ward’s (1994) multi-state models were an improvement to the prior dichotomous studies and both have produced excellent results. A model that can provide additional information to all stakeholders is of immense value. However, these models also have some drawbacks as will be outlined below. An inspection of the two sets of models indicates that different “States of Distress” were used to classify unhealthy companies and only a single state for non-failed companies. The limitation of this approach is that non-failed companies would also have different “States of Health” and an incorrect grouping of the different “States of Health” into one state whilst at the same time specifying different “States of Distress” could result in an inefficient or skewed model being developed.

Lukhwareni (2005)

The Lukhwareni’s (2005) study is a fresh approach to analysing companies and provides good theoretical insight and guidance into issues possibly faced by companies in each of the eight identified stages. In addition, the study 54 would appeal to sector specific investors with the primary aim of investing in companies that show wealth creation and sustainable growth. However, the study as a whole has some drawbacks. Each of the stages (or clusters) of the study, separated companies into above or below sector average companies in each of three variables (Turnover, Operating Income and Return on Capital Employed). One of the drawbacks of the study was that by using absolute values for Turnover and Operating income, the matrix would be biased towards the larger Turnover and Operating Income companies and thereby, penalise the medium to small ones.

MULTIVARIATE DISCRIMINANT ANALYSIS (MDA)

Multiple Discriminant Analysis (MDA) is both a well known and a generally accepted statistical method employed by researchers for comparative studies. According to Balcaen & Ooghe (2006), Altman (1968) introduced the technique to business failure prediction and the former refer to an excerpt from Altman (1968:592) on describing MDA: “A statistical technique used to classify an observation into one of several a priori groups dependent upon the observation’s individual characteristics… [it] attempts to derive a linear [or quadratic] combination of these characteristics which ‘best’ discriminates between the groups”.

STATISTICAL SOFTWARE

The statistical package used for ANOVA and the development of the MDA and CHAID models is SPSS version 13. Originally, it was intended to use Ordinal Logistic Regression (OLR). However, a limitation of the SPSS software was that it was unable to perform Stepwise Ordinal Logistic Regression. In the “Literature Review”, it was pointed out that both MDA and OLR yielded similar results and considering that MDA was being used in this study, CHAID was used as a substitute for OLR.

READ Small'Businesses'Access'to'Finance

DISTRESS – Negative Earnings (State –2)

Failure to react quickly to a dynamic and ever changing business environment and/or a continuum of negative REG will ultimately lead to financial distress namely, Negative Earnings (State –2). It can be argued that one should look for at least two successive negative Earnings prior to considering the company as being in Distress. However, a single year of negative Earnings could be adequate to adversely affect the company’s share price and in the worst case scenario, threaten its survival. This study regards even a single loss as a sign of Distress in that year.

TABLE OF CONTENTS :

DECLARATION
ACKNOWLEDGEMENTS
EXECUTIVE SUMMARY
LIST OF ABREVIATIONS USED
THE DEFINITION OF TERMS
CHAPTER 1 – INTRODUCTION
- 1.1 BACKGROUND
- 1.2 PROBLEM STATEMENT
- 1.3 IMPORTANCE OF THE PROBLEM
- 1.4 RESEARCH OBJECTIVES AND METHODOLOGY
- 1.5 LIMITATIONS OF THE RESEARCH
- 1.6 ASSUMPTIONS MADE IN THE RESEARCH
- 1.7 STRUCTURE OF THE RESEARCH REPORT
CHAPTER 2 – LITERATURE REVIEW
- 2.1 DICHOTOMOUS MODELS (OVERSEAS)
- 2.2 DICHOTOMOUS MODELS (SOUTH AFRICA)
- 2.3 MULTI-STATE MODELS (OVERSEAS)
- 2.4 MULTI-STATE MODEL (SOUTH AFRICA)
- 2.5 CRITICAL ANALYSIS OF PRIOR RESEARCH
CHAPTER 3 – STATISTICAL TECHNIQUES
- 3.1 UNIVARIATE ANALYSIS FOR THE NAÏVE MODEL
- 3.2 MULTIVARIATE DISCRIMINANT ANALYSIS (MDA)
- 3.3 CHI-SQUARE AUTOMATIC INTERACTION DETECTION (CHAID)
- 3.4 STATISTICAL SOFTWARE
CHAPTER 4 – IDENTIFICATION OF THE STATES OF HEALTH
- 4.1 DISTRESS – NEGATIVE EARNINGS (STATE –2)
- 4.2 INTERMEDIATE – NEGATIVE REAL EARNINGS GROWTH (STATE –1)
- 4.3 HEALTHY – POSITIVE REAL EARNINGS GROWTH (STATE 0)
CHAPTER 5 – VARIABLE SELECTION
CHAPTER 6 – MODEL DEVELOPMENT (3 YEAR AVERAGE)
- 6.1 DEVELOPMENT OF TEST AND HOLDOUT SAMPLES
- 6.2 YEAR N MODELS
- 6.3 YEAR N-1MODELS
- 6.4 YEAR N-2 MODELS
- 6.5 YEAR N-3 MODELS
- 6.6 DE LA REY COMPARISON
- 6.7 ANALYSIS OF MODELS
CHAPTER 7 – MODEL DEVELOPMENT (5 YEAR AVERAGE)
- 7.1 DEVELOPMENT OF TEST AND HOLDOUT SAMPLES
- 7.2 YEAR N MODELS
- 7.3 YEAR N-1MODELS
- 7.4 YEAR N-2 MODELS
- 7.5 YEAR N-3 MODELS
- 7.6 DE LA REY COMPARISON
- 7.7 ANALYSIS OF MODELS
CHAPTER 8 – FINANCIAL RISK ANALYSIS MODEL (FRAM)
- 8.1 VARIABLES OF INTEREST
- 8.2 FRAM CUT-OFF POINTS
- 8.3 ANALYSIS OF HOLDOUT SAMPLE COMPANIES
- 8.4 GENERAL OBSERVATIONS
CHAPTER 9 – SUMMARY AND CONCLUSIONS