University of Wollongong. Research Online

Transcription

1 University of Wollongong Research Online University of Wollongong Thesis Collection University of Wollongong Thesis Collections 2007 Evaluation of subjective risk tolerance categorisation methods used by financial advisors: evidence from a psychometric financial risk tolerance questionnaire Zaffar Subedar zaffar@uow.edu.au Recommended Citation Subedar, Zaffar, Evaluation of subjective risk tolerance categorisation methods used by financial advisors: evidence from a psychometric financial risk tolerance questionnaire, PhD thesis, School of Accounting & Finance, University of Wollongong, Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library: research-pubs@uow.edu.au

2

3 Evaluation of Subjective Risk Tolerance Categorisation Methods Used by Financial Advisors: Evidence from a Psychometric Financial Risk Tolerance Questionnaire A thesis submitted in fulfilment of the requirements for the award of the degree of DOCTOR OF PHILOSOPHY from UNIVERSITY OF WOLLONGONG By Zaffar Subedar, B.Com (Hons) School of Accounting and Finance November, 2007

4 Thesis Certification CERTIFICATION I, Zaffar Subedar, declare that this thesis, submitted in fulfilment of the requirements for the award of Doctor of Philosophy, in the School of Accounting and Finance, University of Wollongong, is wholly my own work unless otherwise referenced or acknowledged. The document has not been submitted for qualifications at any other academic institution.... (Signature) Zaffar Subedar 23 November 2

5 Acknowledgements I gratefully acknowledge the support and assistance of my initial principal supervisor, Associate Professor Michael McCrae, for his words of wisdom, objectivity and constant patience. I thank Dr. Dionigi Gerace for coming on board as principal supervisor further into my candidature, adding encouragement and constantly propelling me to completion. I would also like to thank Geoff Davy of FinMetirca Ltd for making the data available, which has made my thesis unique. Heartfelt thanks to Amelia Smith for her love, words of encouragement, home-baked biscuits and cakes, keeping me fit and reading multiple drafts of my entire thesis for grammatical misdemeanours. Sincere thanks go to my friends and colleagues, Joel Grant, Steve Tulig, Sandra Chapple and Ully Silaen for listening and supporting me on campus. Thanks also to Brian Murphy, Andre Dekker, Brendan Burrows and Roger Montgomery for their industry insight and advice on my thesis topic. Special thanks to my close friends Paul Lim and Ben Cleary for understanding me not being able to be the most sociable friend over the last couple of years. Finally, I extend my most earnest gratitude to my family, whose incessant encouragement and confidence in me throughout my candidature kept me on track till the end. 3

6 Table of Contents LIST OF TABLES... 8 LIST OF FIGURES PREFACE ABSTRACT CHAPTER 1: INTRODUCTION TESTING THE EFFECTIVENESS OF FINANCIAL ADVISOR S HEURISTICS IN CATEGORISING INVESTORS RISK TOLERANCES A COMPARISON BETWEEN FINANCIAL ADVISORS HEURISTIC RISK TOLERANCE CATEGORISATION AND INVESTORS ACTUAL CATEGORISED RISK TOLERANCES DIRECT QUESTIONS AND THE EFFECTIVENESS OF FINANCIAL ADVISORS SUBJECTIVE CATEGORISED RISK TOLERANCES SUMMARY CHAPTER 2: RELATED LITERATURE DEFINITION OF RISK AND FINANCIAL RISK TOLERANCE Definition of Risk Definition of Financial Risk Tolerance MEASUREMENT OF RISK TOLERANCE Measuring Risk Tolerance from Portfolio Composition Interview Method of Measuring Risk Tolerance Questionnaire Method of Measuring Risk Tolerance Desirable Characteristics of Questionnaires HEURISTIC ESTIMATES OF RISK TOLERANCE

7 2.3.1 Definition of Heuristics Commonly Used Heuristics for Financial Risk Tolerance Evaluation Age Gender Net Wealth Income Marital Status Education Occupation Other Heuristics RISK TOLERANCE CATEGORISATION Origins of Risk Tolerance Categorisation Previous Studies on Categorised Financial Risk Tolerance Summary of Risk Tolerance Categorisation DIRECT (OBJECTIVE) QUESTIONING FOR RISK TOLERANCE MEASUREMENT Direct Questions in Risk Tolerance Measurement Qualitative Direct Questions Quantitative Direct Questions Risk Preference Reversal Summary of Direct Questioning SUMMARY CHAPTER 3: TESTING THE EFFECTIVENESS OF FINANCIAL ADVISORS HEURISTICS IN CATEGORISING INVESTORS RISK TOLERANCES INTRODUCTION DATA AND SAMPLE METHOD AND EMPIRICAL RESULTS

8 3.3.1 Model Specification Dependent Variables Independent Variables RESULTS Discriminant Analysis Test Results Group Means, Medians and Standard Deviations of Classifying Variables Determination of Differentiating Variables Classification Results CONCLUSION CHAPTER 4: A COMPARISON BETWEEN FINANCIAL ADVISORS HEURISTIC RISK TOLERANCE CATEGORISATION AND INVESTORS ACTUAL CATEGORISED RISK TOLERANCES INTRODUCTION DATA AND METHOD Heuristic (Subjective) Categorised Risk Tolerance Direct Questioning Qualitative and Quantitative Direct Questions RESULTS AND DISCUSSION Severity of Misclassification Implications of Misclassification of Risk Tolerance CONCLUSION

9 CHAPTER 5: DIRECT QUESTIONS AND THE EFFECTIVENESS OF FINANCIAL ADVISORS SUBJECTIVE CATEGORISED RISK TOLERANCES INTRODUCTION DATA AND METHOD RESULTS AND DISCUSSION Difference between Subjective Forms of Risk Tolerance Categorisation General Logistic Results Broad Categorisation Results Narrow Categorisation Results Direct Questions Quantitative Direct Questions CONCLUSION CHAPTER 6: CONCLUSIONS REFERENCES APPENDIX

10 List of Tables Table 3-1 Descriptive Statistics of Sample 95 Table 3-2 Operational Definitions of Categorisation Table 3-3 Test of Equality of Group Means.104 Table 3-4 Group Means, Medians and Standard Deviations of Classifying Variables..107 Table 3-5 Standardised Canonical Discriminant Function Coefficients..111 Table 3-6 Fisher s Linear Discriminant Functions Coefficients..114 Table 3-7 Classification Results Table 4-1 Table 4-2 Scoring for Heuristic Index Frequency Table of Financial Advisors Heuristic Risk Tolerance Categorisation under Broad Categorisation.135 Table 4-3 Frequency Table of Financial Advisors Heuristic Risk Tolerance Categorisation under Narrow Categorisation..136 Table 4-4 Summary of Risk Tolerance Categorisation Movements between Heuristic and Direct Questioning Methods.147 Table 5-1 Table 5-2 Wilcoxon Matched Pairs Test Results.156 Broad categorisation 159 8

11 Table 5-3 Table 5-4 Narrow categorisation..161 Direct Questions

12 List of Figures Figure 2-1 Figure 3-1 Figure 4-1 Risk Tolerance and Portfolio Theory.31 Self-Assessed Risk Tolerance Question 99 Movement between Risk Tolerance Categories (Qualitative Question One)..138 Figure 4-2 Movement between Risk Tolerance Categories (Qualitative Question Two).140 Figure 4-3 Movement between Risk Tolerance Categories (Quantitative Question One)..142 Figure 4-4 Movement between Risk Tolerance Categories (Qualitative Question Two).143 Figure 5-1 Figure 5-2 Broad Categorisation of Subjective Risk Tolerance Methods.162 Narrow Categorisation of Subjective Risk Tolerance Methods

13 Preface Some of the work presented in this thesis is published as joint work in refereed journals and conference proceedings. Chapter 2 has sections from a paper published as: McCrae, M., and Z.Subedar, 2005, Risk Attitude Profiles of Investors in the Financial Services Industry: The Impact of Behavioural Biases, In Poloucek, S, Stavarek, D (eds), Future of Banking after the Year 2000 in the World and in the Czech Republic (Vol X Finance & Banking). Karvina: Silesian University, p ISBN Chapter 4 is currently under review with the Journal of European Financial Management Association. 11

14 Abstract This thesis examines the effectiveness of Financial Advisors subjective categorised risk tolerance assessment methods. Research in this field is scarce and unequivocal, with little guidance on best practice for professionals in the industry. This research provides empirical evidence on an increasingly important issue, given recent regulatory changes in Australia and abroad. Data used in this thesis is drawn from two sections of a psychometrically validated risk tolerance assessment tool. Prior studies examining issues regarding financial risk tolerance use relatively small samples, students, hypothetical investors or respondents of large-scale surveys that cover several financial and lifestyle aspects of individuals. The data set used in this thesis is advantageous as it overcomes these undesirable characteristics, with the majority of respondents being real investors seeking financial advice. The first issue examined is the effectiveness of Financial Advisors use of subjective risk tolerance categorisation (heuristic) methods in statistically differentiating categories of risk tolerance. Heuristics are rules of thumb or short cuts used when making complex decisions. The existing literature examining the directional relationships of demographic characteristics to predict an investor s risk tolerance provides the basis for selecting Financial Advisors heuristics. Results, while similar to previous studies, indicate investors can be classified into four and five categories of risk tolerance, as opposed to 12

15 three in previous research. This finding illustrates that Financial Advisors heuristics are approximations that are not statistically effective as the level of categorisation increases. The second issue examined compares Financial Advisors subjective (heuristic) categorised risk tolerances to an investor s actual categorised risk tolerances. Limited prior research demonstrates substantial variation in Financial Advisor s interpretation of hypothetical investor statements. Thus, if Financial Advisors heuristics are effective in categorising an investor s risk tolerance, no difference between an investor s responses to direct questions and Financial Advisors categorised heuristic risk tolerances should exist. To determine if there are differences, various forms of categorisation are used for comparison. Using two sections of a psychometrically validated risk tolerance assessment instrument, Financial Advisors categorised heuristic risk tolerances are constructed and compared to investors actual responses to direct questions. A statistically significant difference is found between Financial Advisors subjective (heuristic) categorised risk tolerances and investors actual categorised risk tolerances, consistent under the various forms of categorisation. The third issue examined is the likelihood of an investor having the same categorised subjective risk tolerance as their response to direct risk tolerance questions. Investors responses are compared to two measures of subjective categorised risk tolerances using multivariate analysis established in the literature. The two measures are (i) Financial Advisors subjective categorised risk tolerance (as established in chapter four) and, (ii) investors self-assessed categorised risk tolerances (as established in chapter three). Results suggest that the type of direct question is an influencing factor in the 13

16 misclassification of investors risk tolerance categorisation. Quantitative direct-type questions result in investors being more risk-tolerant when facing choices that are quantitative, as opposed to qualitative, in nature. Also, as the level of categorisation increases, the effectiveness of Financial Advisors subjective risk tolerance assessments is reduced. This thesis empirically illustrates three important results regarding the effectiveness of Financial Advisors subjective categorised risk tolerance assessment methods. First, Financial Advisors heuristics are not effective as the level of risk tolerance categorisation increases. Second, in comparison to direct objective measures, there is a significant difference between Financial Advisors subjective categorised risk tolerances and investors responses to direct questions. Third, using different direct question types, Financial Advisors subjective methods become ineffective as the level of categorisation increases. Results of this thesis provide Financial Advisors, investors and regulators with an understanding of the complex nature of risk tolerance categorisation. Further, these findings support a move towards formal risk tolerance categorisation procedures that are effective under several levels of categorisation. 14