EXPANSION PLANNING AT A PROMINENT CANADIAN INVESTMENT MANAGEMENT FIRM by Kelly Anne Chung BSc. in Statistics, University of British Columbia, 2001 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES FACULTY OF COMMERCE AND BUSINESS ADMINISTRATION We accept this thesis as conforming to ^ttie required standard THE UNIVERSITY OF BRITISH COLUMBIA December 5, 2002 © Kelly Anne Chung, 2002 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the r e q u i r e m e n t s f o r an advanced degree a t the U n i v e r s i t y o f B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y purposes may be g r a n t e d by the head o f my department o r by h i s o r her r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t copying or p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l gain s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . Department o f dom^^l The U n i v e r s i t y o f B r i t i s h Columbia Vancouver, Canada Date http://www.library.ubc.ca/spcoll/thesauth.html 2003-03-27 ABSTRACT A prominent financial investment firm currently operates offices located in four major cities across Canada. In addition to their current markets, they are seeking to increase their market share of private investors in a number of Canadian cities over a 10-year planning horizon. Expanding their operations to service these new markets would mean either increasing the capacity of their current offices or acquiring additional offices located in new markets, both of which are costly. Thus, a critical decision must be made as to the most cost effective method of expanding their operations to service new markets. Though costs are expected to be greater for a new office location, it is believed that servicing clients from an office located within the prospective city will positively affect revenue growth for the firm. This assumption along with other factors believed to impact revenue growth, such as positive press coverage, are assessed to determine and quantify their potential impact. As advisors' compensation and support are a large component of the costs incurred to service new markets, a hiring strategy of advisors must also be developed for each new market entered. Results from both our scenario analysis tool as well as our analysis on factors affecting revenue growth have given management the information needed to convince shareholders to invest in expansion planning. Please note that all absolute numbers have been altered within this thesis to adhere to a confidentiality agreement with the financial management firm. T A B L E OF CONTENTS ABSTRACT LIST OF FIGURES LIST OF TABLES ACKNOWLEDGEMENTS 1. INTRODUCTION 1.1. Company Background 1.2. Nature of The Mutual Fund Business in Canada 1.3. Private Client Structure 1.4. Proj ect Obj ective 1.5. Decisions to Be Investigated 1.6. Cash Flow Structure 1.7. Revenue Structure 1.8. Cost Structure 1.9. Literature Review : 2. ANALYSIS METHODOLOGY 2.1. Solution Approach 2.2. Cumulative Net Contribution Growth Models 2.3. Location Effect on Growth of Cumulative Net Contributions 2.4. Marketing Effect on Growth of Cumulative Net Contributions 2.5. Advisor Skill Level Effect on Growth of Cumulative Net Contributions 2.6. Financial Press Coverage Effect on Average Net Contributions 2.7. Proximity Effect on Average Net Contributions 2.8. Cost Model...: 2.9. Profitability Model and Scenario Analysis Tool 3. RESULTS 3.1. Cumulative Net Contribution Growth Models 3.2. Location Effect on Growth of Cumulative Net Contributions 3.3. Marketing Effect on Growth of Cumulative Net Contributions 3.4. Advisor Skill Level Effect on Growth of Cumulative Net Contributions 3.5. Financial Press Coverage Effect on Average Net Contributions 3.6. Proximity Effect on Average Net Contributions 3.7. Profitability Model and Scenario Analysis Tool 4. INTERPRETATION OF RESULTS 5. SUMMARY 6. FUTURE DIRECTIONS AND COMMENTS REFERENCES APPENDIX A PROCESS MAPS: CLIENT ACQUISITION APPENDIX B PROFIT MODEL PARAMETERS : ii iv v vi :. 1 1 1 2 3 3 4 5 5 5 7 7 8 14 16 18 19 20 21 23 25 25 26 31 34 37 47 52 55 56 56 58 59 66 iii LIST O F FIGURES Figure 1.3.1 The Private Investors Structure 3 Figure 1.6.2 Expected Net Cash Flow for a 10 year Expansion Plan '. 4 Figure 1.6.3 Expected Cumulative Net Cash Flow for a 10 year Expansion Plan 4 Figure 2.2.1 All Non-Discretionary Advisors Cumulative Net Contributions '. 10 Figure 2.2.2 All Modelled Non-Discretionary Advisors Cumulative Net Contributions 10 Figure 2.2.4 All Modelled Discretionary Advisors Cumulative Net Contributions. 11 Figure 3.1.1 Non-Discretionary Linear Fit of Cumulative Net Contribution For ALL Advisors 25 Figure 3.1.2 Discretionary Non-Linear Fit of Cumulative Net Contribution For ALL Advisors 26 Figure 3.2.1 Non-Discretionary Cumulative Net Contribution Linear Fits by Location 27 Figure 3.2.2 Normal Cumulative Distribution vs. Residuals Cumulative Distribution 29 Figure 3.2.3 Reisduals vs. Predicted Values 29 Figure 3.2.4 Discretionary Cumulative Net Contribution Non-Linear Fits by Location 30 Figure 3.3.1 Non-Discretionary Advisors Grouped by Marketing Ability 31 Figure 3.3.2 Discretionary Advisors Grouped by Marketing Ability 33 Figure 3.4.1 Non-Discretionary Advisors Grouped by Skill Level 34 Figure 3.4.2 Discretionary Advisors Grouped by Skill Level 36 Figure 3.5.1 Average Net Contributions For a Vancouver Non-Discretionary Advisor 38 Figure 3.5.2 Average Net Contributions For a Toronto Non-Discretionary Advisor 39 Figure 3.5.3 Average Net Contributions For a Calgary Non-Discretionary Advisor 40 Figure 3.5.4 Average Net Contributions For a Vancouver Discretionary Advisor 41 Figure 3.5.5 Average Net Contributions For a Toronto Discretionary Advisor 42 Figure 3.5.6 Average Net Contributions For a Calgary Discretionary Advisor 43 Figure 3.5.7 Unassigned Vancouver Clients 44 Figure 3.5.8 Unassigned Toronto Clients 45 Figure 3.5.9 Unassigned Calgary Clients 46 Figure 3.5.10 Unassigned Other Clients 47 Figure 3.6.1 Average Net Contribution Per Non-Discretionary Advisor For Calgary Clients.... 48 Figure 3.6.2 Residuals vs. Predicted values 49 Figure 3.6.3 Normal Probability Plot of Residuals 49 Figure 3.6.4 Average Net Contribution Per Discretionary Advisor For Calgary Clients 50 Figure 3.6.5 Residuals vs. Predicted Values 51 Figure 3.6.6 Normal Probability Plot of Residuals 51 Figure 3.7.1 Sensitivity Analysis On Maximized Internal Rate of Return 52 Figure 3.7.2 Sensitivity Analysis On Time to Breakeven 53 Figure 3.7.3 Sensitivity Analysis On Maximum Quarterly Investment 54 iv LIST OF TABLES Table 1.2.1 Mutual Fund Assets Under Management, December 2001 2 Table 2.2.1 Transactional File Sample Data 8 Table 2.2.2 Client Location Sample Data 9 Table 2.6.1 Positive Press Events on the Firm 20 Table 2.8.1 Direct Advisor Costs 22 Table 2.8.2 Costs of Opening a New Branch Office 22 Table 3.2.1 Linear Regression Results 27 Table 3.2.2 t-test For Differences in Slope Between Toronto and Calgary advisors 28 Table 3.2.3 Parameter Estimates 30 Table 3.2.4 Wald's Test 30 Table 3.3.3 Parameter Estimates 32 Table 3.3.4 Parameter Estimates 33 Table 3.4.3 Parameter Estimates 35 Table 3.4.4 Test Differences in Slope Between Highly Skilled and Medium Skilled Advisors. 35 Table 3.4.5 Skill Level Parameter Estimates For Discretionary Advisors 36 Table 3.4.6 Walds Test For differences between Highly Skill and Medium Skilled Advisors... 37 Table 3.5.1 T-tests with Equal Variance For Vancouver Non-Discretionary Advisors 38 Table 3.5.2 T-tests with Equal Variance For Toronto Non-Discretionary Advisors 39 Table 3.5.3 T-tests with Equal Variance For Calgary Non-Discretionary Advisors 40 Table 3.5.4 T-tests with Equal Variance For Vancouver Discretionary Advisors 41 Table 3.5.5 T-tests with Equal Variance For Toronto Discretionary Advisors 42 Table 3.5.6 T-tests with Unequal Variance For Calgary Discretionary Advisors 43 Table 3.5.7 T-tests of Publications Effect on Vancouver Client's Net Contributions 44 Table 3.5.8 T-tests of Publications Effect on Toronto Client's Net Contributions 45 Table 3.5.9 T-tests of Publications Effect on Calgary Client's Net Contributions 46 Table 3.5.10 T-tests of Publications Effect on Other Client's Net Contributions 47 Table 3.6.1 T-test for Average Net Contributions Per Non-Discretionary Advisor 48 Table 3.6.2 T-test for Average Net Contributions Per Discretionary Advisor 50 v ACKNOWLEDGEMENTS I would like to thank members of the project team: Jason Goto, Martin Puterman, and Glen Darou for their support throughout the project and for giving me the opportunity to work on this business problem. I would also like to thank my sponsors at the financial management firm, who brought forth a challenging, and interesting business problem for me to tackle. Firstly, I would like to recognize all my fellow classmates for their support and selfless sharing of knowledge during our program. In particularly, I would like to express my gratitude to my friend Anita Parkinson for her friendship, encouragement, and insight throughout my project. A special thanks to my friend Earl Gordon, for lending a hand during a stressful period, and to Jonathan Patrick, for taking the time to work and discuss new ideas with me. I would also like to express my gratitude to my family who have always believed in my ability; no matter what my endeavour. Lastly, I would like to thank my Academic Advisor, Martin Puterman for his insight and technical guidance into my thesis; and my Second Reader, Anming Zhang for taking the time to review my work and give feedback. Thanks to everyone for making this a great learning experience. vi 1. INTRODUCTION 1.1. Company Background The research for this study was conducted at a prominent Canadian investment management firm with over $30 billion in Assets Under Management and four office locations across Canada. The firm produces it's own suite of mutual funds and provides advisory services to non-profit organizations, institutional clients and private investors. In the early years, the firm solely managed institutional pension plans but has been expanding to provide services to private investors. Their main method of gaining new business is through word of mouth from satisfied clients. Their funds' performances have consistently ranked in thefirstquartile with notable mentionsfromleaders in the industry. Today, thefirmis known for its long-term performance record, low client fees, and professional counselors. 1.2. Nature of The Mutual Fund Business in Canada The mutual fund industry was the fastest-growing segment of the financial services sector during the 1990s, with assets increasingfrom$25 billion in December 1990 to $426 billion by December 2001. For almost every year since 1990, the majority of asset growth has been due to sales. The main types of mutual funds are money market funds, bond funds, equity funds, dividend funds, mortgage funds and real estate funds. In 2001 there were about 1,800 mutual funds in Canada representing over 50 million unit-holder accounts. The industry is fairly consolidated. From a total of 80 fund management companies operating in Canada, the top 10 companies account for 72 per cent of all assets. At the end of 2001, there were 144 dealerfirmsinvolved in the sale of mutual funds and a total of 90,000 people employed in the industry. The industry can generally be divided into the producers and the distributors of mutual funds, although banks and credit unions are involved in both sides of the industry. Companies involved solely in the production of mutual funds tend to distribute their funds through investment advisors,financialplanners, mutual fund dealers and life insurance agents, while the banks and credit unions distribute their fund products through their retail branches as well as through brokers. Producers and distributors earn revenue from three main categories of fees: management fees (to pay for the management of the fund), transaction fees (to pay for buying and selling the fund units) and special fees (for specific administrative costs). The domestic mutual fund industry is experiencing increasing competitionfromlarge foreign firms, with three foreign fund companies (AIM Funds Management Inc., Fidelity Investments Canada Limited and Franklin Templeton Investments Corp.) now managing 20 per cent of mutual fund assets in Canada. Industry analysts expect the mutual fund industry to continue to consolidate as competition intensifies and the manufacturers of mutual funds seek to reduce costs. Table 1.2.1 shows the top 10 companies, in terms of assets, and their percentage of Canadian market share. 1 Table 1.2.1 Mutual Fund Assets Under Management, December 2001 Company Assets Market Share ($ millions) (percent) Investors Group Inc. 41,644 9.8 Royal Mutual Funds Inc. 36,571 8.6 34,704 8.1 Mackenzie Financial Corporation 33,280 7.8 Fidelity Investments Canada Limited 32,590 7.6 TD Asset Management Inc. 31,925 7.5 AGF Management Limited 28,235 6.6 CIBC Securities Inc. 26,344 6.2 Cl Mutual Funds Inc. 21,175 5.0 Franklin Templeton Investments Corp. 19,711 4.6 Total 306,179 71.8 Industry total 426,398 100.0 1 AIM Funds Management Inc. 1 Investors Group Inc. acquired Mackenzie Financial Corporation in April 2001. However, they remain separate and distinct in their distribution, their investment management operations and their brands. Source: IFIC. 1 The mutual fund industry in Canada is currently governed by provincial and territorial legislation, which regulate the underwriting, distribution and sale of securities. There is also extensive self-regulation by the Investment Dealers Association of Canada, the Mutual Fund Dealers Association of Canada and the stock exchanges. 1.3. Private Client Structure Though thefirmservices non-profit organizations, institutional clients and private investors, our project focuses solely on the Private Client business. The Private Client services are divided into two main offerings: Discretionary and Non-Discretionary services. Discretionary service provides a customer with a one-on-one relationship with a Discretionary Advisor who will make investment decisions on behalf of the client, under specific investment objects determined by the client. A client who opts for Discretionary services has the option to invest solely on the firm's mutual funds or in a combination of funds and securities to further customize his or her portfolio. However, a client must meet a minimum dollar investment level before choosing to invest in a combination of funds and securities. Non-Discretionary service offers clients several service offerings: Assigned, Unassigned, and Nominee. Assigned clients have a one-on-one relationship with a Non-Discretionary advisor. Unassigned clients (House accounts) do not have a relationship with a Non-Discretionary advisor 2 but invest in the firm's mutual funds directly with the firm. Nominee clients are clients who invest in the firm's mutual funds but do not purchase them through thefirmand instead acquire them through other brokers or dealers. Figure 1.3.1 illustrates the Private Client Structure within the firm. Figure 1.3.1 The Private Investors Structure Discretionary ilp|ipMi J Mutual Funds and Securities 1.4. I Non-Discretionary "I .1 Project Objective The objective of this project was to develop a decision support tool for management to enable them to evaluate profit margins of various strategic growth plans over a 10-year planning horizon. This tool would allow management to determine the impact of opening an office in a given city and hiring additional advisors throughout a 10-year period. Our sponsor expressed that this study would be deemed a success if we could "provide management with a usable, scalable, and easy to understand tool" that would enable them to carry out analysis on their own, long after the study terminated. 1.5. Decisions to Be Investigated The following decisions were to be investigated in our study: a) Should a new market be serviced out of an existing office or through a new office within the market? b) How many advisors should be hired and when shall they be hired? c) Is there an effect on revenue growth due to advisor skill level and marketing ability, and can we quantify this effect? d) Is there an effect on revenue growth for Discretionary and Non-Discretionary advisors due to positive press coverage on the firm? e) Is there an effect on revenue growth from being present (through a local office) in a new market compared to servicing the new market with advisors who travel to see their clients? If so, can this be quantified? 3 1.6. Cash Flow Structure For each expansion, the firm undertakes must meet specific requirements set by their board of directors. The main specification is that a 10-year expansion plan should not result in a maximum negative cash flow of x amount for any quarter of the 10 years. This constrained the number of possible expansion plans according to the value of x. As net cash flow is generated from revenue minus costs, an example of the expected cash flow for each month throughout a 10 year planning horizon in Figure 1.3.2. Figure 1.6.2 Expected Net Cash Flow for a 10 year Expansion Plan H — Total Revenue — Total Direct and < > £)'< X X )<X X ¥r*( X Indirect Costs <XX XX X Xj .Net Cash Row 12 16 20 24 28 36 32 40 T i m e (quarters) Here, we see the expected revenues for each quarter along with expected costs for each quarter, and the resulting net cash flows at each quarter. This graph, however, does not show the time ' when the investment into this expansion plan will breakeven. Thus, Figure 1.6.3 shows the Cumulative Net Cash Flows for each month of the planning horizon. Figure 1.6.3 Expected Cumulative Net Cash Flow for a 10 year Expansion Plan 0 4 8 12 16 20 24 28 32 36 40 T i m e (quarters) 4 In the previous graph, we can see that the investment will break even when the cumulative net cash flow is 0. Thus the investment is expected to become profitable after this point in time. 1.7. Revenue Structure Although revenue in the mutual fund business is generally composed of management fees, transaction fees, and special administrative fees, the firm's only source of revenue is through management fees. A client's management fee is a percentage of the amount they have invested with the firm. Percentages differ depending on the type of mutual fund the client is investing in. In general, funds containing international securities will have higher fees than funds containing domestic funds. Thus, a client's fee will be a weighted average of the dollar amount invested in a mutual fund and the percentage fee associated with that fund. The firm's management fees have been and continue to be amongst the lowest in the industry. 1.8. Cost Structure Costs to the firm are generally made up of Advisor's compensation, maintaining trading systems cost, processing transactions cost, and general administrative costs. However, during the company's expansion process, they incur additional costs such as new office costs, which must be modelled when projecting expected future cash flow. This will be discussed further in section 2. 1.9. Literature Review The topic of capacity expansion has been widely researched for organizations, which have substantial expansion costs with long payout times, and which benefit largely from economies of scale. As our problem lies in the service industry, such expansion costs are generally much smaller and the benefits from economies of scale are considerably less. Traditional capacity expansion problems also tend to focus on the manufacturing or utility industries, which have considerable transportation costs associated with the expansion of their operations. I refer to Shulman (1991) who has developed an algorithm for solving dynamic capacity plant location problems with discrete expansion sizes. Other researchers have focused mainly on static location problem formulations. I refer to Efroymson and Ray (1966), and Geoffrion and McBride (1978). Bienstock and Shapiro (1988) who have developed a stochastic program with recourse to optimize resource acquisition decisions in the electric utility industry. In this paper, Bienstock and Shapiro evaluate acquisitions based on two criteria, how the acquisition will benefit the company over the long-term, and how the acquisition will affect the company in terms of the uncertainty that the new acquisition will create. Thus, they propose a stochastic program with recourse to model possible contingency plans to the realization of each acquisition. In the service industry, transportation costs are minimal for making expansion decision, while other issues such as meeting fluctuations in demand are more difficult to deal with due to the service industry's lack of ability to store inventories to meet these fluctuations. I refer to Berman, Ganz, and Wagner (1994) who have formulated a stochastic optimization model for 5 planning capacity expansion for a service industry, which incorporates uncertainty of future demand. They develop a weighted set of possible demand scenarios in a stochastic optimization model, which determines a schedule for expanding capacity. The solution provides the size, location and timing of each expansion to maximize the companies expected profit. Using Lagrangian relaxation and taking advantage of the special nested knapsack structure of the subproblems, they developed an algorithm to solve this specific problem. We opted for a simpler model to assist management, as pursuing a stochastic optimization model would force the firm to hire additional employees and new software to support the model. We developed a scenario tool in Microsoft Excel, which would simulate possible expansion plans for a given city. This would allow transparency and usability to the model for persons without the knowledge of stochastic optimization models. 6 2. ANALYSIS M E T H O D O L O G Y 2.1. Solution Approach We began the study by discussing potential differences between the Discretionary and NonDiscretionary services. This allowed us to determine the feasibility of building a profitability analysis model incorporating both businesses. We analyzed differences in cost, revenue and business processes for maintaining each level of service. This was accomplished by working with the Chief Financial Officer, the Vice Presidents of the Discretionary and Non-Discretionary services, and an Associate in Discretionary Services. Through interviews with several Discretionary and Non-Discretionary Advisors, process maps were developed to better understand theflowof clients into the firm. Flow charts are included in appendix A. Once potential differences were highlighted, we determined that differences among the groups did not warrant creating separate profitability model for each type of service. We began by developing models to estimate expected revenue growth and costs projected over a 10-year horizon. We then sought to quantify assumptions made by upper management regarding the impact of factors believed to affected revenue growth. Thus, factors such as positive press coverage, and proximity of the advisor with respect to their clients were analyzed. Once these parameters were estimated, they could, in combination with the revenue and cost model, be incorporated into a profitability model, which would be developed in Microsoft Excel. This particular platform was chosen due to its presence in thefirm,its transparency, and its userfriendly logic. The following outlines the major outcomes of this study. 1. Process Maps - Process maps were developed for the acquisition and maintenance of Discretionary and Non-Discretionary Advisors 2. Expected revenue growth models - separate models were developed for the Discretionary and Non-Discretionary advisors to reflect different growth natures 3. Analysis on factors affecting revenue growth - advisor skill level and marketing ability, positive press coverage, and proximity effect were analyzed to quantify their effect on revenue growth 4. Expected Cost model - separate models were developed for the Discretionary and NonDiscretionary advisors based on figures from thefirmsActivity Based Costing analysis 5. Expected Profitability Model - a profitability model was developed to reflect the expect cash flows over a 10-year horizon by incorporating the revenue and cost models as well as parameters for effects of advisor skill, marketing ability, positive press coverage, and proximity effect. 6. Scenario Analysis Tool - a scenario analysis tool automated the profitability model to analyze hundreds of expansion strategies, allow the user to optimize on one of several criteria and receive sensitivity analysis on the optimal expansion plan 7 2.2. Cumulative Net Contribution Growth Models As revenue is a percentage of the accumulated Net Contributions/Book Value for each advisor, we needed to model Book Value growth per advisor in order to obtain estimates of revenue. This was quite challenging, as the firm had never previously queried this data from their systems. The firm currently only keeps reports on an advisor's current Assets Under Management. Thus, they do not keep historical reports on either Assets Under Management or Book Value for each advisor. Therefore, we looked to their transactional records to create data on historical advisor Book Value for our analysis. Analysts from the firm queried two flat files from their data warehouse, one pertaining to the historical transactions of an investor and the other to the current residential address of the investor . Data was only available from October 1998 to May 2002. Table 2.2.1 and 2.2.2 shows a sample of the transactionalfileand the investor locationfileused in our analysis. Table 2.2.2 shows a sample of the resulting flatfileafter processing the data for use in our models. 7 Table 2.2.1 Transactional File Sample Data Market Value Dealer Code Dealer R e p Code Dealer Count 234,567 300 12 1 1,624,357 -10,000 300 12 1 1,623,333 1,000 300 12 1 Investor N o Date 1123 082000 1,234,567 1123 092000 1123 102000 C a s h Flow a) Investor No — A unique investor number is assigned for each client. This number is used in processing transactions on their accounts with the firm. b) Date - The month and year when the transactions took place. c) Market Value - The current market value of the portfolio. Market value is also known as Assets Under Management. This value incorporates the contributions by the investor, switches between accounts, as well as the market growth of the portfolio d) Cash Flow - The net amount of cash into the portfolio. Total buys minus total sells of all accounts within a portfolio during a month. e) Dealer Code - The financial institution in which the client makes the transaction. f) Dealer Rep Code - The Advisor who carries out the transaction. g) Dealer Count - The number of Advisors who have made transactions for the client in a given month. 1 Current data was used in this study as no historical client locations were kept in their data warehouse. 8 Table 2.2.2 Client Location Sample Data Investor N o City Province C o d e Postal C o d e Country Code 1123 Toronto ON L5M4J2 Canada 1124 Vancouver BC V6T4M3 Canada 1125 Calgary AB T2L4M8 Canada a) Investor No-A unique investor number is assigned for each client. This number is used to make transactions on their many accounts with the firm. b) City - The client's current mailing address city as of May 2002 c) Province Code - The client's current mailing address province as of May 2002 d) Postal Code - The client's current mailing address postal code as of May 2002 e) Country Code - The client's current mailing address country as of May 2002 A new file was created by merging the client transaction file with the client location file. A new variable, Cumulative Net Contributions, was calculated in order to model revenue growth. A definition is provided below as well as a description of how it was created. a) Cumulative Net Contribution (CNC) - The total accumulated Net Contributions of a client's portfolio aggregated by advisors from when they began with the firm. As data is only , availablefromOctober 1998, Net Contributions of advisors who started working with the firm before October 1998 needed to be estimated. This was accomplished by taking a percentage of the Assets Under Management at October 1998 to represent the Cumulative Net Contributions up until this date. Since the growth of an advisor's Book Value was dependent on the time the advisor was hired, a time variable was created to represent the amount of time the advisor had been with the firm, in months. Thus, each advisors growth pattern of Cumulative Net Contribution was plotted against their time with the firm in order to determine whether their existed similar patterns in growth amongst the advisors. From initial plots of Cumulative Net Contribution for the Non-Discretionary and Discretionary advisors, it was observed that the Non-Discretionary advisors growth pattern followed a linear shape, whereas the Discretionary advisors followed a non-linear shape. This distinction prompted us to develop separate growth models for the Non-Discretionary and Discretionary advisors. In our analysis, several discretionary and non-discretionary advisors were identified as outliers due to their unusual growth. They were eliminated from our analysis as further reconciliation with members of the firm validated our observation. It was found that these advisors either were part time advisors or were retiring and thus, had been giving away large amounts of their Assets Under Management to other advisors in the firm. 9 All Non-Discretionary advisors Cumulative Net Contributions growth curves are shown in Figure 2.2.1. All Discretionary advisors who were used in fitting Cumulative Net Contributions growth are shown in Figure 2.2.2. All Discretionary advisors Cumulative Net Contributions growth curves are shown in Figure 2.2.3. Similarly, all Discretionary advisors who were used in fitting Cumulative Net Contributions growth are shown in Figure 2.2.4. The legend on the righthand- side shows each advisors growth. Each advisor is labelled with a number to protect the advisors confidentiality. Figure 2.2.1 All Non-Discretionary Advisors Cumulative Net Contributions -*—35 _o—36 37 27 -X—25 ^>—32 c o .a E c o -0—28 o 29 ^—79 -•—39 26 _^31 30 —i 33 tf ***** OJ > E o -x— 34 ggf_— CM CD CO CN OO •tf o co CN •tf co CO CD 00 o o CN CN CO Time with the Firm (Months) Figure 2.2.2 All Modelled Non-Discretionary Advisors Cumulative Net Contributions .35 .36 c o .37 "3 -27 .a '£ c o O .25 .32 .28 CD .29 z a> > .79 .39 E o 26 — CN CnrrrftjS tf CN CO CO co •tf o CO CN •tf co CD CO o o CN CN CO Time with the Firm (Months) 10 Figure 2.2.3 All Discretionary Advisors Cumulative Net Contributions Non-Discretionary Advisors As the Non-Discretionary advisor's growth followed a linear pattern, a simple linear regression model was used to model growth of Cumulative Net Contribution with a single independent variable, amount of time the advisor has been with the firm as an advisor. No intercept was included into the model as it is customary for an advisor to begin the growth of their book with no clients. Equation (2.2.1) shows the linear model used to estimate Non-Discretionary book growth. Y =fi x+ e (2.2.1) x Where, Y = Cumulative Net Contribution x = Amount of time the Advisor has spent growing their book (months) Pi = Effect on Cumulative Net Contribution due to amount of time the advisor is with the firm 8 = Unexplained variance For each Advisor, we obtained monthly totals of Cumulative Net Contributions for the each month the advisor has been working with the firm. As data was only available from October 1998, to May 2002, the number of observations used for each advisor in the model was dependant on when the advisor was hired. Therefore, approximately 30 data points for each advisor was used in the linear regression model. In order tofitthe model we treated each advisor data point independently form one another. The regression model was created in SAS, as statistical software program, using each of the monthly Cumulative Net Contributions for every advisor independently. The following hypothesis test was performed to determine whether x was a significant predictor of Y. A result was noted as significant if its P-value based on a t-test was less than 0.05. Discretionary Advisors As with the Non-Discretionary advisors, Cumulative Net Contributions for each advisor, on a monthly basis was calculated and used to model Book Value growth. For the same reasons, noted for the Non-Discretionary advisors, only approximately 30 data points were used in the modelling of Discretionary advisors. Every Discretionary advisors monthly data that was available was used to fit the model. As the growth of Discretionary advisor Book Value did not follow a linear pattern, and instead showed a characteristic Sigmoidal growth curve, a Sigmoidal Non-Linear growth model was used to model the Discretionary advisors book growth. We tested thefitsof several 3 and 4 parameter, non-linear growth models and chose to model book growth using a four parameter Non-linear model as it allowed greater flexibility to thefitcompared to a 3 parameter model. More specifically a 4-parameter, Weibull model was used tofitthe growth curve. The Weibull model function is included in (2.2.2). Y = a-0exp(-jx ) + e S (2.2.2) 12 Where, Y = Cumulative Net Contribution / Book Value x = Amount of time the Advisor has spent growing their book (months) a = Asymptote a - P = intercept on the Y-axis (initial value of Y) Y = rate at which the response changes from its "initial value" 5 = rate at which the response changes from its "initial value" e = Unexplained variance The following steps outline how a fit was developed: 1. Obtain initial parameter estimates of a, P, y, 5 2. Obtain convergence of parameter estimates 3. Fit the model Obtain Initial Parameter Estimates Initial parameter estimates of a, P are obtained by graphing Y versus x. cto is obtained by estimating the approximate maximum value approached by the response Y as X approaches infinity. For low values of x, an estimate of the ordinate intercept Y I T can be estimated with a value corresponding to x=0. From cto and YINT, an estimate of Po is obtained as N 0= 0~ INT S a <--> Y 2 2 3 By rearranging (2.3) we obtained, -log a = log^ + <JlogX r (2.2.4) 0 J J Performing a simple linear regression of Zo on log X enables estimates of log y and 8o. Obtain Convergence of Parameter Estimates Once initial estimates are obtained, the Gauss-Newton method can be used to find convergence of parameter estimates. The Gauss-Newton method is not included in this document but can be found in the Ratkowsky (1983). Fit the model Once parameters have converged, the model can be estimated. Thefinalestimated parameters for the model using all the data were: a = 183,000,000 log y = -5.899303 0= 183,000,000 5 = 1.553743 13 All estimations and tests are performed in, SAS, a statistical software package using the 'proc model' statement. In addition to the effect of time on the growth of an advisors book, Management was also interested in the effects of location, advisor skill and marketing ability. In the next sections we outline the methodology used to test for these effects. As there are very few advisors, one large model could not be used to determine the interaction effects of the variables. 2.3. Location Effect on Growth of Cumulative Net Contributions Management wanted to determine whether there were significant differences in the book growth of advisors from different office locations. To test differences among advisors from the Vancouver, Toronto, and Calgary offices, linear and non-linear models, with indicator variables, were created for the Discretionary and Non-Discretionary Advisors, respectively. These indicator variables allow the model to distinguish the data by location. Non-Discretionary Advisors To test for location effects on Cumulative Net Contribution growth, we expand model (2.2.1) to include indicator variables for location. The proposed linear model is shown in (2.2.5). As noted in section 2.2, no intercept is included in the model as it is assumed that an advisor begins with no clients. Y = p x + P (a *x) l 2 x + P (a *x) 3 2 +z (2.2:5) where, Y = Cumulative Net Contribution / Book Value x = Amount of time the Advisor has spent growing their book (months) [0, Vancouver Advisor ai = [l, Toronto Advisor [0, Vancouver Advisor a = [l, Calgary Advisor 2 Pi = Effect on Cumulative Net Contribution due to amount of time with the firm 02 = Effect on Cumulative Net Contribution due to Toronto Location 03 = Effect on Cumulative Net Contribution due to Calgary Location s = Unexplained variance The following hypothesis tests were performed to determine significant differences in revenue growth among advisors from each location. Ho: 02 0, Ha: 02 * 0, Ho: 03 = 0, = Toronto advisors growth rates do not differ from Vancouver advisors Toronto advisors growth rates do differ from Vancouver advisors Calgary advisors growth rates do not differ from Vancouver advisors 14 Ha: P3 * 0, Calgary advisors growth rates do differ from Vancouver advisors T-tests were performed and significance of theses hypotheses was determined from the resulting P-valuesfromWald's test. If the resulting P-value is less than 0.05, the null hypothesis would not be rejected. Ho: p2 = p3, Toronto advisors do not differ from Calgary advisors Ha: P2 * P3, Toronto advisors do differfromCalgary advisors Residual analysis was then conducted on the final model to ensure the assumptions of normality and independence are not violated. Discretionary Advisors To test for differences in Book Value growth among advisorsfromdifferent locations, a fourparameter Weibull model, with indicator variables on parameter y, is proposed. Normal business practices require a Discretionary advisor to reach approximate capacity at 229,000,000 Assets Under Management over a 5-year period. However, as our model focuses on Book Value we need to define a capacity constraint on Book Value growth. As Assets Under Management is defined as Cumulative Net Contributions with added market growth, plus transferred assets from other advisor, we proposed a capacity constraint of 183,000,000 for Book Value growth. Our proposed value was also in line with management's expectations. Thus, this value was used to estimate the asymptote variable, a. As an advisor normally begins with an empty book, the initial starting parameter, a - P, is set equal to 0. We hold a and p constant while we fit parameters y, and 8. Ideally, we would want to test changes in a (the maximum value) as well as y (the initial rate of change from the starting value). However, there did not exist enough data, near to the asymptotic value, to model. This was shown in the results of our preliminary fitted models, which estimated a well over 213,000,000. As this was the case, we propose to model changes in y only. Thus, (2.6) shows the Weibull model tested in our analysis. Y = 183,000,000-183,000,000exp(-y + 0ft +0 a ) x + e S 2 2 (2.2.6) Where, Y = Cumulative Net Contribution / Book Value x = Amount of time the Advisor has spent growing their book (months) 0, Vancouver Advisor 1, Toronto Advisor 0, Vancouver Advisor [1, Calgary Advisor Pi = Effect on initial growth rate, parameter y, due to Toronto Location P2 = Effect on initial growth rate, parameter y, due to Calgary Location e = Unexplained variance 15 The following hypotheses tests were performed to determine statistical significance among initial revenue growth rate of advisors at each office location. Ho: Pi = 0, Toronto advisors growth rates do not differ from Vancouver advisors Ha: Pi ^ 0, Toronto advisors growth rates do differ from Vancouver advisors Ho: 02 = 0, Ha: p * 0, 2 Galgary advisors growth rates do not differ from Vancouver advisors Calgary advisors growth rates do differ from Vancouver advisors Appropriate T-tests were performed and statistical significance was concluded if the P-value resulting from the T-test was less than 0.05. Parameter estimates were also analyzed to determine magnitude of effect. Ho: 0i = 02, Toronto advisors growth rates do not differ from Calgary advisors Ha: 0i * 02, Toronto advisors growth rates do differ from Calgary advisors Similar to the Non-Discretionary analysis, Wald's test statistic was used to determine significance of each hypothesis. 2.4. Marketing Effect on Growth of Cumulative Net Contributions The firm wanted to determine whether advisors who proactively marketed for new clients showed an increased rate of growth in Net Contributions compared to advisors who did not actively seek out new clients. The firm's management separated their advisors into 2 categories of marketing ability: Reactive marketer and Proactive marketer. Reactive marketers generally do not make cold calls or conduct information sessions. They generally acquire new clients from their existing clients, other advisors, or through the firm's contact centre, where general inquiries are handled. One the other hand, a Proactive advisor is known to actively seek our new clients, through contacts with accountants and other business professionals; information sessions, and cold calling. A similar methodology was used to test for differences in advisor growth rate between the two marketing styles as was used in section 2.3. Non-Discretionary Advisors To test for differences in growth rates among Non-Discretionary Advisors, a linear regression model with indicator variables was used to test for effect of proactive marketing. The proposed linear model is an extension of model 2.1 and is shown in (2.7). Y = P x + P mx + e x 2 (2.2.7) Where, Y = Cumulative Net Contribution / Book Value x = Amount of time the Advisor has spent growing their book (months) 16 0, Reactive Advisor mi = < [\, Proactive Advisor P i = Effect on Cumulative Net Contribution due to the amount of time the advisor is with the firm P 2 = Effect on Cumulative Net Contribution due to proactive marketing 8 = Unexplained variance The following tests were created to assess for significant differences among the Proactive and Reactive advisors. Ho: P2 = 0, Ha: P2 * 0, Proactive marketing does not affect Book Value growth Proactive marketing does affect Book Value growth T-tests were performed and statistical significance was concluded if the P-value resulting from the T-test was less than 0.05. Parameter estimates were also analyzed to determine magnitude of the effect. Residual analysis was not conducted on model (2.2.7) and should be conducted in further research. Discretionary Advisors To test whether proactive marketing activity affected the growth rate of advisor Book Value, we performed T-tests on the parameter y in the Weibull curve. Thereby determining the significance on the initial rate of Book Value growth between the two marketing styles. The Weibull model used in this analysis is shown in (2.2.8). Y = 183,000,000 -183,000,000 exp(-^ + A m.) x + e 5 (2.2.8) Where, Y = Cumulative Net Contribution / Book Value x = Amount of time the Advisor has spent growing their book (months) f 0, Reactive advisor mi= < _ [1, Proactive advisor Pi = Effect on parameter y due to proactive marketing 8 = Unexplained variance The following test was performed to determine whether proactive marketing had a significant effect on the initial growth rate, y, of advisor Book Value. Ho: Pi = 0, Ha: Pi * 0, Proactive marketing does not affect on initial Book Value growth rate Proactive marketing does affect on initial Book Value growth rate Statistical significance was concluded if the P-value resultingfromthe T-test was less than 0.05. Parameter estimates were also analyzed to determine magnitude of effect of proactive marketing. 17 2.5. Advisor Skill Level Effect on Growth of Cumulative Net Contributions The firm's management sought to quantify differences in Book Value growth among advisors who, in their opinion, possessed different skill levels. Management used subjective judgment to categorize their advisors into three categories for Skill (0=Low Skill level, l=Medium Skill level, 2=High Skill level). Low Skilled advisors are advisors who have performed worse, in comparison to their peers, in the past, or advisors who have not been with the firm long enough for their ability to be determined. Medium Skilled advisors describe those who have performed well, however, inconsistently in the past High Skilled advisors describe those who have performed above average on a consistent basis, in comparison to their peers. Non-Discretionary Advisors For the Non-Discretionary advisors, we used a linear model with indicator variables to categorize advisors by skill level. We then tested for significance between advisor skill levels. As advisors generally begin growing their book with no initial clients, no intercept parameter is needed in the model. Thus, interaction effects are only included to test for differences in book growth among different skilled advisors. The model is shown in (2.2.9). Y = P x + P s x + PiS x + s x 2 x 2 (2.2.9) Where, Y = Cumulative Net Contribution / Book Value x = Amount of time the advisor has spent growing their book (months) f 0, Low Skilled advisor Sl= \ [1, High Skilled advisor f 0, Low Skilled advisor s= < [\, Medium Skilled advisor 2 0 i — Effect on Cumulative Net Contribution due to the amount of time with the firm 02 = Effect on Cumulative Net Contribution due to being High Skilled 03 = Effect on Cumulative Net Contribution due to being Medium Skilled £ = Unexplained variance The following tests were performed to assess for significant differences among the rated advisors. Wald's test statistic was used to determine the significance of each hypothesis. Ho: p = o, Ha: 02*0, 2 High Skilled advisors growth rate does not differ from low skilled advisors High Skilled advisors growth rate does differ from low skilled advisors Ho: p = o, Medium Skilled advisors growth rate does not differ from low skilled advisors Ha: 03*0, Medium Skilled advisors growth rate does differ from low skilled advisors 3 Ho: 02 = 03, High Skilled advisors growth rate does not differ from medium skilled advisors Ha: 02*03, High Skilled advisors growth rate does differ from medium skilled advisors 18 Discretionary Advisors For the Discretionary advisors, non-linear Weibull models, with indicator variables, were used to model the effect of skill level on initial rate of growth of Cumulative Net Contributions. Thus, only tests on parameter y are conducted. The model used in our analysis is shown in (2.2.10). Y = 183,000,000 -183,000,000exp(-y + ^sl + ^sl) x+ e S (2.2.10) Where, Y = Cumulative Net Contribution / Book Value x = Amount of time the Advisor has spent growing their book (months) f 0, Low Skilled Advisor S l _ [l, High Skilled Advisor [0, Low Skilled Advisor [1, Medium Skilled Advisor Pi = Effect on parameter y due to a High Skilled advisor P2 = Effect on parameter y due to a Medium Skilled advisor s = Unexplained variance We performed the following tests to compare High and medium skilled advisors to Low advisors and then compared high to medium skilled advisors. Ho: Pi = 0, Ha: Pi * 0, High Skilled advisors growth rate does not differ from low skilled advisors High Skilled advisors growth rate does differfromlow skilled advisors Ho: P2 = 0, Ha: P2 * 0, Medium Skilled advisors growth rate does not differfromlow skilled advisors Medium Skilled advisors growth rate does differfromlow skilled advisors The P-valuesfromT-tests were used to determine significance of these hypotheses. Parameter estimates were also analyzed to determine magnitude of effect of high and medium skill level. Ho: Pi = P2, High Skilled advisors growth rate does not differfrommedium skilled advisors Ha: Pi * P2, High Skilled advisors growth rate does differfrommedium skilled advisors Wald's test of significance was used to determine differences in initial growth rate between high and medium rated advisors. Statistical significance was concluded if the P-value resulting from a T-test was less than 0.05. 2.6. Financial Press Coverage Effect on Average Net Contributions Management wanted to quantify the effect of positive coverage in financial publications on the amount of Net Contributions accumulated by each advisor. Table 2.6.1 shows the list of events, 19 which were analyzed. Analysis was broken down into Discretionary and Non-Discretionary clients (Assigned and Unassigned). Assigned clients were then further broken down by location of advisor (Vancouver, Toronto, or Calgary). Unassigned clients were however, broken down by client location (Vancouver, Toronto, or Calgary). Event 1 refers to coverage in a Canadian Mutual Funds buyer guide published yearly. Event 2 corresponds to a notable mention in a nationally distributed Canadian newspaper and Events3 and 4 refers to articles in a nationally distributed Canadianfinancialmagazine. Table 2.6.1 Positive Press Events on the Firm Event Date Positive Event 1 November 1999 Positive Event 2 January 2001 Positive Event 3 November 2001 Positive Event 4 D e c e m b e r 2001 Note: As Event 3 and event 4 occur relatively close to each other, they are modelled as 1 event. As new advisors were hired between these events, our analysis focused on Average Net Contributions per advisor to eliminate the effect an additional new advisor would have on the event in question. Therefore, Net Contributions were calculated for each advisor for each month from the period between October 1998 and May 2002. T-tests were performed and based on the resulting P-values a determination was made as to whether there was a change in Average Net Contributions one and a half years before and six months following the event. In cases where there were not enough data points, prior to the event, to obtain a one and a half year sample, we simply used the data that was available for our sample. Since Non-Discretionary clients do not have a relationship with an advisor, Average Net Contributions per advisor could not be calculated. Instead, Total Net Contributions of all clients were calculated and assessed to determine changes before and after an event. As with the Discretionary analysis, data one and a half years prior to the event and six months after the event was used for analysis. T-tests were performed and used to determine statistically significant changes in total Net Contributions. The Average Net Contribution per client was not analyzed, as the statistic would not have accounted for new clients joining the firm. 2.7. Proximity Effect on Average Net Contributions Determining whether physical presence in a given market affected the average accumulated Net Contributions for each advisor was critical, as one of our main decisions involved whether to open an office in a new market. Therefore, a trade off between the extra cost of the new office and expected revenue growthfromthat new office, had to exist if we were to recommend opening an office in the new market. Determining whether the physical presence in the new market, through a local office, influenced the growth of Average Net Contributions per advisor for both the Discretionary and Non20 Discretionary services was key. We compared the Average Net Contributions of the Calgary clients before and after opening the Calgary office. This office was chosen because data on advisor growth was available before and after the opening of the office, in November 1999. For the Discretionary analysis, the firm's management had four Vancouver advisors assigned to seek out Calgary clients prior to opening the Calgary office, although these advisors were also responsible for managing Vancouver clients. Thus, each assigned advisor represented the growth rate of a part time Vancouver advisor devoted to managing Calgary clients. Thus, only the four Discretionary advisors were compared and analyzed against the Calgary advisors. Their Average Net Contributions were calculated for each month between October 1998 and May 2002. As Discretionary managers have a non-linear growth curve, we would expect a non-linear increase in Net Contributions following the opening of the Calgary office, due to the new hired advisors. This presented a challenge, as the assumption of constant variance was violated. To adjust for this, we performed a two-sample t-test with unequal variance, using Satterthwaite's approximation (Steel and Torrie, 1980) for calculating the degrees of freedom. For Non-Discretionary advisors, there is no problem with non-constant variance and therefore a t-test was performed to test for changes in Average Net Contributions before and after the new office. One must note that the Non-Discretionary advisors were not specifically assigned to acquire Calgary clients. Residual analysis was then conducted on the final model to ensure assumptions of normality and independence were not violated. 2.8. Cost Model Data for modelling cost was obtained from the firms Activity Based Costing analysis. The Activity Based Costs were separated into Direct Advisor Costs, Indirect Advisor Costs, and Fixed Business Costs. Direct costs were defined as costs that could be directly traced back to each advisor. Indirect costs were defined as costs such as processing transactions, and administrative costs, which could not be directly traced back to each advisor, but were instead divided among all the advisors. Fixed costs were defined as costs, which are incurred regardless of the number of advisors hired. For example, the costs of researching investments for the firms mutual funds remain the same whether they hire 1 or 10 discretionary advisors. Table 2.8.1 to 2.8.3 lists the Activity Based Costs in each category. Aside from the normal business costs, costs for opening a new branch office were also estimated. These costs are outlined in Table 2.8.4. 21 Table 2.8.1 Direct Advisor Costs Direct A d v i s o r C o s t s Indirect A d v i s o r C o s t s Fixed B u s i n e s s C o s t s Advisor Salary M a i n t a i n client s y s t e m s Economic Research A d v i s o r benefits Maintain P C s R e s e a r c h i n g Investments Assistant salary Maintain the Network W e b Site A s s i s t a n t benefits Professional Memberships Corporate Communication (Travel c o s t s ) Mail a n d Office S u p p l i e s Maintaining Strategy Courier and Postage Maintaining C o m p l i a n c e Trading Losses Maintaining P e o p l e Registrations, L i c e n c e s , a n d F e e s Vintage Fund Processing Costs MMF Fees Overall Corporate Administration Other Funds Trading and Settlements Activity Based Costs were already divided between Discretionary and Non-Discretionary therefore only per advisor costs were needed. The method used to calculate Direct advisor cost, Indirect advisor costs, and Fixed costs was tofirstseparate costs into the appropriate categories. This was conducted in consultation with the firm's Chief Financial Officer. Once these costs were categorized, a number of full time employees was needed to determine a per advisor cost. This was available in the firm's activity based costing analysis. As the firm's activity based costs were still under revision, some adjustments were made to allocate the appropriate proportion of costs to the Discretionary and Non-Discretionary advisors, which were not accomplished in the Activity Based Costing analysis. Table 2.8.2 Costs of Opening a New Branch Office Office Set-up Costs Rent and operating Alarm/Internet Hiring and Business Planning Premise Insurance Receptionist The Office Set-up costs include furniture, signage, and other one-time expenses. Other ongoings costs were estimated such as rent, and ongoing premise insurance. 22 2.9. Profitability Model and Scenario Analysis Tool A profitability model was constructed in MS Excel and incorporated the cost model as well as the calculation of revenue growth. Once Book Value growth was created, a percentage of Book Value at each month estimated expected revenue streams. The profitability model was based on quarterly profit figures, as this is the norm for analyzing financial data. The model was built to incorporate many parameters such as inflation rate for costs and market growth rate for estimating the amount of Assets Under Management for each advisor. The firm had conducted previous market analysis on the cities in which they were considering expanding to and, thus, knew the size of each potential market in terms of Assets Under Management. Therefore, our model incorporated a parameter for adjusting the size of each new market as well as a parameter for changing the percentage of market share the firm was aiming to capture in a 10-year horizon. In addition to the parameters included in the model, additional controls were added to define whether a new office would be placed in a new market. This would affect costs for servicing the new market, as there would be additional office set up costs. Other controls were added to define whether a new advisor would be hired and when in the 10-year horizon this would occur. Currently the model can include hiring 3 Discretionary and 3 Non-Discretionary advisors, although, this could easily be scaled to include more advisors. Once all parameters are entered into the model, graphs as well as scenario statistics are generated to allow the user to understand the profitability margin of the growth plan. The model outputs the following statistics when a scenario is entered: Net Present Value, Internal Rate of Return, total Assets Under Management after 5-years, Total Assets Under Management after 10-years, Time when the investment will breakeven, and Maximum Quarterly Cash outlay during the 10 year investment. The graphs that are created for each scenario are 10-year Revenue growth curves, 10-year expected costs, and 10-year net profit streams. The main decision is to determine the most profitable 10-year plan based on 3 decisions: 1. Whether to open an office in the new market or whether to service the new market through advisors from their existing office? 2. How many Non-Discretionary advisors to hire and when to hire them? 3. How many Discretionary advisors to hire and when to hire them? A VBA script was written to automate the manual process of considering every combination of the above 3 decision. The script will look at every scenario possible, by changing the parameters associated with the above 3 decision and store the scenario statistics in a large file. Once the statistics of every scenario has been gathered. The user will have the opportunity to choose the growth plan, which maximizes one of 4 criteria: Internal Rate of Return, Net Present Value, Total Assets Under Management after 10 years, and Index of Investment. 2 Index of investment is defined as the ratio of the net present value of an investment and the maximum investment. 23 In addition, the model also outputs sensitivity analysis for the "optimal" expansion plan. This report explains the impact of specific changes to assumptions or variables on the profitability of the optimal expansion plan. We looked at the following factor changes in our sensitivity analysis: the addition or subtraction of a Discretionary advisor, the addition of a NonDiscretionary Advisor, changes in market growth rate, changes in delaying hiring an advisor, and changes to whether to service the new market locally, from a new office or remotely, from an existing office. 24 3. RESULTS 3.1. Cumulative Net Contribution Growth Models Figure 3.1.1 and Figure 3.1.2 show every Non-Discretionary and Discretionary advisor's Cumulative Net Contributions for each monthfromthe time they began advising with the firm. Calgary advisors are marked with a triangle, Toronto advisors are marked with a cross, and Vancouver advisors are marked with a solid line. From the above graph we see that there is a Calgary advisor who is clearly outperforming the other advisors. As thefirmdid not view this advisor different in anyway from the other advisors the advisor was kept in the model. Thefinallinear regression model is shown in (3.1.1). It is expected that an advisor will accumulate approximately $399,981/month in Net Contributions. A Y =399,9S\TimeWithFirm (3.1.1) As the advisors growth rates are quite variable, the fitted line does not capture the large variance in Book Value as the amount of time the advisor is with thefirmincreases. From the graph, the Book Value of advisors who have been working for thefirmfor 2 years rangefroma high of $37 Million to a low of $10 Million. The range in Book Value at 2 Vi years ranges from a high of $53 Million to a low of $14 Million. The standard error of the predicted value at 2 years is approximately $2.1 Million. We would like to address if similarities are present among advisors with similar growth rates. In the next section we will test the effects of Location, Advisor's Marketing ability, and Advisor's Skill level. 25 Discretionary Advisors As noted in Section 2.1, the Weibull fit for modelling Discretionary advisors growth is restricted to $183,000,000 as there is insufficient data to constrain the model from exceeding regular business practices of $229,019,748 in Assets Under Management. Figure 3.1.2 Discretionary Non-Linear Fit of Cumulative Net Contribution For ALL Advisors ^ ^ 3 * X C o O a> T ' / y c >u E o 4 6 7 .8 9 10 11 .16 17 18 19 20 21 22 23 All Fit CM o CM Time with the Firm (Months) The final 4-parameter Weibull model, using datafromall advisor locations is shown in (3.1.2). Y = 183,000,000-183,000,000 *exp(-5.899303)A: 1 553743 (3.1.2) From the graph we can see a lot of variability around the fitted line. The Book Value of advisors working with the firm for 2 years, rangefrom$16 Million to $106 Million. The Weibull curve estimates a book of 2 years to be valued at $61 Million. As with the Non-Discretionary advisors, we will test for other factors, which could be contributing to the variation around book growth. After 5 years of building their book, a Discretionary advisor is estimated to acquire approximately $145 Million of client Net Contributions. 3.2. Location Effect on Growth of Cumulative Net Contributions Figure 3.2.1 and Figure 3.2.2 show each Non-Discretionary and Discretionary Advisor's Cumulative Net Contribution growth against the amount of time the advisor has been with the firm. Calgary advisors are marked with a triangle, Toronto advisors are marked with a cross, and Vancouver advisors are marked with a solid line. 26 Non-Discretionary Advisors Linear fitted lines for each advisor location are also shown for Vancouver, Toronto, and Calgary in addition to each advisors actual book growth. Figure 3.2.1 Non-Discretionary Cumulative Net Contribution Linear Fits by Location Time with the Firm (Months) Notice from the above graph that there are only two Calgary advisors to model. The fit of the Calgary advisors show that at the end of a 3-year period, they will accumulate approximately $17 Million. We can see that the fitted line is fitting advisor 25, as advisor 26 has very few data points, which may be misleading if advisor 25 is an exceptionally skilled advisor. We can also see that Vancouver and Toronto's fitted growth curve are somewhat similar, with Vancouver advisors expected to accumulate approximately $14 Million, and Toronto advisors expected to accumulate approximately $9 Million after a period of 3 years. Table 3.2.1 shows the results from the linear regression model. Parameter estimates and Pvalues state the significance of each variable in the model. Table 3.2.1 Linear Regression Results Variable DF X 1 a! * x (Toronto) 1 a * x (Calgary) 1 2 Parameter Estimate Standard Error 400,000 8,145 -128,080 15,310 86,600 20,797 t Value Pr>|t| 72.17 -10.11 23.23 <.0001 <.0001 * <.0001 * From the regression results, both Toronto and Calgary advisors book growth are shown to be statistically different to the growth of Vancouver advisors as the calculated P-value is less than 0.05. Vancouver advisors Net Contribution growth rate is estimated at $400,000/month. Toronto advisors are expected to accumulate $128,080 /month less than a Vancouver advisor, at $271,920/month. On the other hand, Calgary advisors are expected to accumulate $86,600/ 27 month more than a Vancouver advisor at $486,600 /month. As the estimates for both Calgary and Toronto advisors were based on very few advisors, we decided to use the fitted Cumulative Net Contribution growth for all the advisors in our profitability model. Thus Toronto advisors, in general, accumulate 32.02% less than Vancouver advisors. Calgary advisors, however, accumulate 21.65% more Net Contributions than Vancouver advisors, on average. Table 3.2.2 shows the resultsfromthe hypothesis test that Toronto advisors growth and Calgary advisors growth do not differ. Table 3.2.2 t-test For Differences in Slope Between Toronto and Calgary Advisors Source DF Mean Square Error F Value Pr>F Pi 1. 3.47E+15 761.85 <.0001 * (3 300 4.56E+12 2 As expectedfromthe results of table 3.2.1, there does appear to be a significant difference in growth rate between Toronto advisors and Calgary advisors. Thefinallinear models for advisor location are shown in (3.2.1). Models for each advisor location are shown in (3.2.2), (3.2.3), and (3.2.4) for Vancouver, Toronto, and Calgary advisors, respectively. A Y = 400,000* -128,080 (^ * JC) + 86,600 (a * *) (3.2.1) A Y = 400,000* (3.2.2) A Y = 271,920* (3.2.3) A Y = 486,600* (3.2.4) Although, our tests show that Calgary advisors tend to grow their books at a faster rate, we must also consider that there are only two advisors data available to model Calgary advisors. This is generally not a large enough sample to perform such statistical tests. We conducted residual analysis on model (3.2.1) to assure that assumptions of normality and independence were not violated. Figure 3.2.2 shows the normal probability plot of the residuals. Figure 3.2.3 shows the residuals plotted against the predicted values. 28 Figure 3.2.2 Normal Cumulative Distribution vs. Residuals Cumulative Distribution mnet_oont = 587825 fsfar rrtR-ISiri rm -1547 58d -M83210e - 1. O O. O O. 1 0.'2 O. 3 O. 4 O. 5 O. 6 O. T O. S O. 9 From this normal probability plot, we see that the residuals are not quite following the normal distribution. The residuals at the tails of the distribution tends to be following that of the normal curve, however, the residuals towards the centre tend to deviate from normal distribution. The differences, however, do not seem to suggest that the residuals are seriously violating the normal distribution. Figure 3.2.3 Reisduals vs. Predicted Values aoooooo eoooooo ra 3 T3 4000000 '55 m 2000000 - 2000000 . + -+ + i- + - eoooooo Predicted Values In this plot, we see non-constant variance and no trends in the residuals. In particular, the variance appears to decrease near the mid-point of the predicted values. 29 Discretionary Advisors Figure 3.2.4 Discretionary Cumulative Net Contribution Non-Linear Fits by Location T i m e with the Firm (Months) Table 3.2.3 shows the resulting parameter estimates from the Weibull curve used to model Location effect. Table 3.2.3 Parameter Estimates Parameter Estimate Standard Error t Value Pr>|t| Y p (Toronto) 6.121407 0.1966 0.04 31.14 <0001 0.48 0.6298 0.0758 2.92 0.0528 30.46 0.0036 * <0001 2 p (Calgary) 5 3 0.019283 0.221214 1.608188 The above P-values show that we cannot reject the hypothesis that initial growth rate of Toronto advisors equals the initial growth rate of Vancouver advisors. However, we can reject the hypothesis that Calgary's advisors initial growth rate is equal to Vancouver's advisors initial growth rate. Table 3.2.4 Wald's Test W a l d ' s Test Statistic Pr> P =P 6.80 0.0091 * 2 3 X 2 Wald's test statistic was used to confirm that Toronto advisors and Calgary advisors have different initial growth rates. With a P-value less than 0.05, we can reject the hypothesis that Toronto and Calgary growth rates are equal. As we could not reject the hypothesis that Toronto 30 advisors' book growth rate was equal to Vancouver advisors' book growth rate, their modelled Weibull Curves remain the same. As Calgary advisors' initial growth rate was found to be greater than that of Vancouver advisors, we expect to see a steeper increase in growth. The final model for Discretionary advisors Book Value growth by location is shown in (3.2.5). Vancouver/ Toronto and Calgary's fitted linear models are shown in (3.2.6) and (3.2.7). Y = 183,000,000-183,000,000exp(-6.1214 + 0.2212 ) a j L 6 0 8 2 X (3.2.5) Y = 183,000,000-183,000,000* exp(-6.1214)* (3.2.6) Y = 183,000,000-183,000,000 *exp(-6.3426)* (3.2.7) L6082 16 0 8 2 Thus, our results showed that after 5 years of building their book, Vancouver and Toronto advisors are estimated to accumulate $145,623,047 Million. Furthermore, a Calgary advisor, with an equivalent 5-year period are estimated to accumulated a Book Valued at $157,815,767 Million. This translated to approximately 10 to 15 more clients per advisor. 3.3. Marketing Effect on Growth of Cumulative Net Contributions Figures 3.3.1 and Figure 3.3.2 show the Non-Discretionary and Discretionary advisors Cumulative Net Contributions for each month they have been working with the firm. They are segmented by marketing ability. Advisors rated proactive are marked with a solid line and advisors rated Reactive are marked with a triangle. Non-Discretionary Advisors Figure 3.3.1 Non-Discretionary Advisors Grouped by Marketing Ability 31 Time with the Firm (Months) From the Figure 3.3.1, we can see that the rate of book growth of Proactive advisors is greater than the rate of growth for Reactive advisors. A Proactive advisor is expected to accumulate approximately $42 Million after 5 years of building their book, whereas a Reactive advisor is expected to accumulate close to $32 Million over the same time period. Thus, a Proactive advisor shows a 31.25% increase in Cumulative Net Contributions compared to a Reactive advisor. Table 3.3.3 Parameter Estimates Standard Error Parameter Estimate Variable DF X 1 700,000 m, * x 1 204,088 t Value Pr > |t| 16,019 32.98 <.0001 22,809 6.75 <.0001 * From this table, we can see that Proactive advisors tend to have a faster rate of growth compared to Reactive advisors. For each month, a Proactive Non-Discretionary advisor will accumulate approximately $204,088 more than an equivalent Reactive advisor. The final linear model is shown in equation (3.3.1). The resulting Cumulative Net Contribution model for proactive and Reactive Non-Discretionary marketers is shown in (3.3.2), and (3.3.3). A Y = 700,000* + 204,088 (m. * *) (3.3.1) A Y = 700,000* + 204,088 * * (3.3.2) A Y = 700,000* (3.3.3) Over a 5-year period, a proactive Non-Discretionary advisor will expect to accumulate $12.2 Million more than a Reactive advisor. This result was very interesting to management as they sought for solid evidence that their subjective assessment of Proactive advisors were correct. They will in the future target advisors with the skills to proactively market for new clients. 32 Discretionary Advisors Figure 3.3.2 Discretionary Advisors Grouped by Marketing Ability t &WYYY\ CM •tf CN CO 00 •tf co o CN CO •tf oo o CO oo o CN Time with the Firm (Months) The above graphs show the Weibull fitted curves for Proactive and Reactive Discretionary advisors. Proactive advisors tend to show an increased rate of growth compared to NonDiscretionary advisors. Table 3.3.4 show the results of our test for differences in initial growth rate between Proactive advisors and Reactive advisors. After 3 years of building their book, a proactive advisor will accumulate $98,012,236 Million compared to a Reactive advisor with $91,999,964 Million. This difference corresponds to approximately 4 additional clients for Proactive advisors. Table 3.3.4 Parameter Estimates Parameter Estimate Standard Error t Value Pr>|t| Y 5.837433 0.2138 27.3 <.0001 5 0.092945 0.0465 2 0.046 Pi (proactive) 1.529785 0.0573 26.7 <.0001 * Wald's test of Proactive effect on Cumulative Net Contribution growth shows that we can reject the hypothesis that Proactive and Reactive advisors initial rate of growth, y, are equal. The final model for Proactive and Reactive advisors is shown in 3.3.5. The model for Proactive advisors is shown in 3.3.6 and for Reactive advisors in 3.3.7. Y= 183,000,000- 183,000,000 * exp(-5.837 +1.529 a,)* ' 0 Y = 183,000,000- 183,000,000 *exp(-5.837 + 1.529)x Y= 183,000,000- 183,000,000 * e x p ( - 5 . 8 3 7 ) x ° - + f 092 a092 092 + (3.3.5) (3.3.6) (3.3.7) 33 After a 5-year period of building a book, a proactive Discretionary advisor is expected to acquire approximately $148,896,418 compared to a Reactive advisor who is expected to acquire $143,393,607. Thus a proactive advisor is estimated to have 3.8% more in Cumulative Net Contributions in comparison to a Reactive advisor, after 5 years. This increase translates to approximately 4 more clients for Proactive advisors than Reactive advisors. The difference between Proactive and Reactive advisors is not as large as management had originally predicted. It was suggested from the firm's management that there could be a spill over effect, where Proactive advisors are passing leads to Reactive advisors, which could explain why the difference in growth of their proactive and Reactive advisors is relatively small. 3.4. Advisor Skill Level Effect on Growth of Cumulative Net Contributions Figures 3.4.1 and Figure 3.4.2 show the Non-Discretionary and Discretionary advisors Cumulative Net Contributions for each month they have been working with the firm. They are segmented into 3 levels for skill: High, Medium, and Low. Advisors rated High Skilled are marked with a solid line, advisors rated Medium Skilled are marked with a cross, and advisors rated Low Skilled are marked with a triangle. Non-Discretionary Advisors Figures 3.4.1 show the Non-Discretionary advisor's fitted linear curves for High, Medium, and Low Skilled advisors. From Figure 3.4.1 we see that Medium Skilled advisors outperformed High Skilled and Low Skilled advisors, which is surprising. We notice that advisor 25 outperformed all other advisors including other High Skilled advisors. Table 3.4.3 show the resultsfroma linear regression model. Table 3.4.3 Parameter Estimates Variable DF Parameter Estimate Standard Error t - Value Pr>|t| X 1 800,000 50896 9.97 <.0001 S i * x (high) S *x(med.) 1 123,088 52577 1.49 0.1388 1 246,991 53602 2.92 0.0038 * 2 From above, we can reject our hypothesis that Medium Skilled advisors have the same rate of growth as Low Skilled advisors. For each month, a Medium Skilled Non-Discretionary advisor is estimated to accumulate $123,088 more than an equivalent Low Skilled advisor. On the other hand, we cannot reject our hypothesis that High Skilled advisors, including the large growth of the outlier (advisor 25) have the same rate of growth of Cumulative Net Contributions as Low Skilled advisors. This finding was very surprising and puzzling to members of the firm. Table 3.4.4 show the resultsfromtesting differences in growth rates among High Skilled and Medium Skilled advisors. Table 3.4.4 Test Differences in Slope Between Highly Skilled and Medium Skilled Advisors Source DF Mean S q u a r e F Value Pr > F Pi 1 9.37E+13 13.53 0.0003 * P2 248 6.93E+12 As the P-value is less than 0.05, we can reject the hypothesis that Medium and High Skilled advisors have the same rate of growth. This concurs with our previous observations that Medium Skilled advisors and Low Skilled advisors show large differences in growth. However, High Skilled advisors and Low Skilled advisors do not show large differences in growth. The final linear model for all skill levels is shown in equation (3.4.1). The model for High and Medium Skilled advisors is shown in equation (3.4.2) and (3.4.3). Y = 800,000* +123,088 (j * *) + 246,991 (s * x) + s (3.4.1) Y = 800,000* + s (3.4.2) Y = 1,046,991 (s x) + e (3.4.3) 2 On average a High and Low Skilled advisors accumulates $ 48,000,000 in a 5-year period, compared to approximately $ 63,000,000 for Medium Skilled advisor. This did not make intuitive sense to members of the firm and thus they considered the possibility that their subjective assessments of their advisors may have been biased. 35 Discretionary Advisors Figure 3.4.2 Discretionary Advisors Grouped by Skill Level From Figure 3.4.2 we see that, as observed with the Non-Discretionary advisors, the Medium Skilled advisors are showing greater growth of Cumulative Net Contributions. From the above graph, we can see that the weibull curves for Low rated advisors show a delayed exponential growth of Net Contributions compared to the growth for High and Medium rated advisors growth, affirming assumptions made by our sponsors. Table 3.4.5 show the results from our tests on parameter y, in the Weibull model proposed in Section 2.5. Table 3.4.5 Skill Level Parameter Estimates For Discretionary Advisors Parameter Estimate Standard Error t - Value Pr > |t| Y 5.944062 0.1705 34.85 <.0001 5 Pi (High) p (Medium) 0.332218 0.042 7.91 <.0001 0.174601 0.0484 3.61 0.0003 * 1.518386 0.0467 32.53 <.0001 * 2 As the resulting P-value is less than 0.05, we can reject our hypothesis that High Skilled Discretionary advisors book growth rate is equal to the book growth rate of Low Skilled advisors. The same can be concluded for Medium Skilled advisors, as we can reject our hypothesis that Medium Skilled advisors book growth rate is equal to the book growth rate of Low Skilled Discretionary advisors. These results affirm assumptions made by the firm's management. 36 Table 3.4.6 Walds Test For differences between Highly Skill and Medium Skilled Advisors Test Statistic Pr > C h i S q u a r e d Label Pi = P 2 11.82 0.0006 H i g h E f f e c t = M e d i u m Effect As the resulting P-value is less than 0.05, we can reject our hypothesis that Medium and High Skilled advisors have the same initial book growth rate. Moreover, Medium Skilled advisors are estimated to have a larger initial growth rate than High Skilled advisors, which conflicts with assumptions made by from the firm. Thefinalmodel is shown in 3.4.4. Models for High, Medium, and Low Skilled Discretionary advisors are shown in 3.4.5, 3.4.6, and 3.4.7 respectively. A Y= 183,000,000- 183,000,000 exp(-5.944 +0.1746^ +1.5184 s )x0.3322 A Y= 183,000,000- 183,000,000exp(-5.944 +0.1746) x 0.3322 A Y A Y (3.4.4) (3.4.5) 183,000,000- 183,000,000exp(-5.944 +1.5184) x 0.3322 (3.4.6) 183,000,000- 183,000,000 exp(-5.944) (3.4.7) 0 3 3 2 2 x After 5 years of growing a book, a High Skilled Discretionary advisor will accumulate approximately $144,681,525 compared to a Medium Skilled advisor who will accumulate approximately $153,687,290 and a Low Skilled advisor with $133,737,661. Thus, a High Skilled advisor will have approximately 8.2% more client Net Contributions than a Low Skilled advisor. A Medium Skilled advisor, however, will acquire approximately 14.8% more than a Low Skilled advisor. 3.5. Financial Press Coverage Effect on Average Net Contributions The timing of positive financial coverage in magazines and newspaper publications were analyzed on Average Net Contributions for a Vancouver, Toronto, and Calgary advisor to determine whether there were significant increases or decreases 6 months after a positive event compared to one and a half years. As noted in Section 2.6, if data is not available to create a one and a half year sample before the event, the data points, which were available, represented the amount of Net Contributions before the publication. We analyzed Non-Discretionary advisors, Discretionary advisors, and Unassigned Clients separately to understand how these events affect different business areas of the private client sector. Figures 3.5.1 to 3.5.3 show Average Net Contributions for Non-Discretionary advisors. Figures 3.5.4 to 3.5.7 show Average Net Contributions for Discretionary advisors. Figures 3.5.8 to 3.5.11 show total Net Contributions for all Vancouver, Toronto, Calgary, and Other Clients as described in Section 2.6. 37 Non-Discretionary Figure 3.5.1 Average Net Contributions For a Vancouver Non-Discretionary Advisor ± Event 1 • Event 2 • Event 3 C O C c o 1 - O D O C C n c O O c n 1 - O D n c T - ) T C C c D n T - N n O c C O O c CN O o O C n D O o CN O O o O C T O o o CM N o C - T o — o CM o O O C o o CM O D 0 T— o o CM O o o 5 CO T— CM o o CM Date (Months) o CM O O CM CD O CM O O CM From Figure 3.5.1 we see that Vancouver advisor acquires larger Average Net Contributions after the positive publication compared to the months prior. Table 3.5.1 summarizes the results of the significance of each event on Vancouver Discretionary advisors. Table 3.5.1 T-tests with Equal Variance For Vancouver Non-Discretionary Advisors Standard Deviation Degrees of Freedom t - Value Pr > Itl 206,429 295,188 12 -2.59 0.0235* 1,061,335 509,883 835,355 345,089 22 -2.21 0.0382 * 1,393,577 370,105 Before Event 3 946,903 374,018 23 -0.02 0.9808 After E v e n t 3 952,963 341,054 Event Before Event 1 After E v e n t 1 Before Event 2 After E v e n t 2 Mean Net Contribution The above tests, allow us to draw several conclusions in regard to the effect of Event 1, Event 2, and Event 3 on the Average Net Contributions per advisor. As the P-values of the first hypotheses tests is less than 0.05, we can reject the hypothesis that the Average Net Contributions prior to Event 1 and following Event 1 are equal. Similar conclusions can be made of Event 2 as the resulting P-value is less than 0.05. We can reject the hypothesis that the Average Net Contributions before and after Event 2 are equal. However, a similar conclusion cannot be made for Event 3 as the resulting P-value is greater than 0.05. Thus we cannot reject the hypothesis that the Average Net Contributions before and after Event 3 are equal 38 Our results show that the Average Net Contribution per advisor was $159,596 per month before and $718,249 per month after Event 1, a 350% increase. Similar growth is found after Event 2, where advisor's Average Net Contributions grew from $570,578 per month before the event to $935,358 per month after the event, a 64% increase. In contrast, Average Net Contributions do not appear to change after the occurrence of Event 3, where advisor's Average Net Contribution increases from $643,471 to $647,431 per month, which represents a mere 0.6% increase. Figure 3.5.2 Average Net Contributions For a Toronto Non-Discretionary Advisor o CO •> T3 < i— CO Q. ^ Event 1 c o + Event 2 CD • Event 3 o CD CD CO V 1 cu > • < CO o CD CD CD CD CD OJ CD o o> o CD CD CD CO O O O O CN CD O O O O CN CD O O O O CN CO o o o O O O CN CN CD O CO O O O CN O O CN B O O CN CO O CN O O CN CO O CN O O CN Date (Months) As Toronto Non-Discretionary advisors were hired one month before Event 1, we cannot compare the Net Contributions 6 months after the event to Net Contributions prior to the event. The Average Net Contributions after Event 2, and 3 are slightly larger than the contributions prior to the events. Table 3.5.2 show the results on the effect of each Event on Net Contributions from clients managed by Toronto Non-Discretionary advisors. Table 3.5.2 T-tests with Equal Variance For Toronto Non-Discretionary Advisors Mean Net Contribution Standard Deviation Degree of Freedom t - Value Pr > Itl 76,462 105,596 6 -1.1 0.3141 After E v e n t 1 354,807 217,065 Before Event 2 364,596 237,377 20 -2.86 0.0096 * After E v e n t 2 991,577 433,852 Before Event 3 686,524 341,966 23 -0.9 0.3766 After E v e n t 3 881,031 229,549 Variable Before Event 1 39 The above results allow us to make the several conclusions regarding the impact of Event 1, Event 2, and Event 3 on the Average Net Contributions of Toronto Non-Discretionary advisors. The P-valuesfromthe test of Event 1 and Event 3 are both greater than 0.05. Thus we cannot reject our hypothesis that the Average Net Contributions before and after Event 1 are equal. Similarly, we cannot reject our hypothesis that the Average Net Contributions before and after Event 3 are equal. On the other hand, we obtain a P-value greater than 0.05 for our test of the effect of Event 2 on Average Net Contribution. Thus, we can reject our hypothesis that the Average Net Contributions before and after Event 2 are equal. The Average Net Contribution following Event 2 changesfrom$262,953 to $672,664 per month, which corresponds to a 156% increase. Figure 3.5.3 Average Net Contributions For a Calgary Non-Discretionary Advisor o <n '> T3 < CO 0c o O CO < CO o OJ OJ OJ CO o OJ OJ OJ OJ o OJ o > OJ CM T— cn OJ OJ CO o o o o CM CO o o o o CM OJ o o o o CM CM CO o o o CM T— o o o CM T— CO O ^— o o CM OJ O CM T O O CM — o o CM CO O CM O O CM CO o CM O O CM Date (Months) The Calgary office was opened in November 1999, therefore, no data is available prior to the Event 1. There does not seem to be any change in Net Contributions following Event 2. There does appear to be large contributions after Event 3. Table 3.5.3 shows the resultsfromthe hypotheses tests conducted for each event. Table 3.5.3 T-tests with Equal Variance For Calgary Non-Discretionary Advisors Variable Before Event 2 After E v e n t 2 Before Event 3 After E v e n t 3 Mean Net Contribution Standard Deviation 909,121 625,076 1,090,000 380,408 893,436 484,337 1,430,000 592,440 Degrees of Freedom t - Value Pr > Itl 18 -0.65 0.5253 23 -2.33 0.029 * 40 The test for the effect of Event 2 on Average Net Contribution results in a P-value that is greater than 0.05. Thus we cannot reject the hypothesis that Calgary advisors' Average Net Contribution before and after Event 2 are equal. Our test for Event 3, however results in a Pvalue, which is less than 0.05, thus we can reject our hypothesis that Calgary advisors' Average Net Contribution before and after Event 3 are equal. Calgary advisors show an increase in Average Net Contributionfrom$893,436 to $1,430,000 per month, which corresponds to a 60% increase. Discretionary Analysis Figure 3.5.4 Average Net Contributions For a Vancouver Discretionary Advisor o CO •> "O < a> D_ o O < O > C O C M I O 0 ) 0 ) C C 0 O O C ) 0 0 ) 0 - a 5 O ) 0 •" .'<— O O O * - 0 C T > M C 0 ) 0 ) ) 0 ) 0 O 0 0 0 i— C O O ) O ) 0 ) 0 C M a O O 5 C T M C 0 - C O O ) 0 0 0 0 0 0 0 0 ) C N C M T - C M O O O O T > O C C i — M 0 0 0 M M 0 C Q O I— 0 0 C C — T— 0 0 C M C M 0 0 C 0 M C O C N 0 0 M C M Date (Months) It appears that the variance of Net Contributions increase dramatically after September 1999 to May 2001, where the variance appears to be stabilizing. From this graph, we cannot observe major changes before and after the three events. Table 3.5.4 show the resultsfroma t-test with equal variance assumption for the Vancouver Discretionary advisors. Table 3.5.4 T-tests with Equal Variance For Vancouver Discretionary Advisors Variable Before Event 1 Mean Net Contribution Standard Deviation 1,160,000 905,384 After Event 1 771,540 1,650,000 Before Event 2 895,020 1,360,000 1,030,000 988,118 772,869 968,429 1,010,000 558,128 After Event 2 Before Event 3 After Event 3 Degrees of Freedom t - Value Pr > Itl 17 0.67 0.5117 22 -0.22 0.8263 23 -0.62 0.5442 41 From these results, we cannot reject our hypotheses that Vancouver advisors Average Net Contributions are the same, before and after each of the 3 favourable publications. It appears that after Event 1 there is a sharp decrease in Net Contributions two months after the event. Both, Events 2 and 3 show increase in Average Net Contributions 6 months following the event. We test hypotheses on Average Net Contributions one and a half years before each event and 6 months after. Table 3.5.5 T-tests with Equal Variance For Toronto Discretionary Advisors Variable Before Event 1 Mean Net Contribution Standard Deviation 1,350,000 782,346 829,710 1,410,000 Before Event 2 1,050,000 916,470 After E v e n t 2 1,550,000 568,666 Before Event 3 1,200,000 601,627 After E v e n t 3 2,140,000 790,296 After E v e n t 1 Degrees of Freedom t - Value Pr > Itl 17 1.05 0.3079 22 -1.26 0.2195 23 -3.19 0.004 * From this summary of results from our t-tests, we cannot reject our hypotheses that the Average Net Contributions before and after Event 1 are the same for Toronto advisors. Similarly, Toronto advisors do not show a significant increase in Average Net Contributions following Event 2. However, there does seem to be a significant effect of Event 3 on the Average Net Contributions for Toronto advisors. As we obtain a P-value, which is less than 0.05, we can 42 reject our hypothesis that the Average Net Contributions for Toronto advisors is not equal before and after Event 3. Prior to the Event 3, a Toronto Discretionary advisor acquired and average of $1,200,000 per month compared to $2,140,000 per month following Event 3. Figure 3.5.6 Average Net Contributions For a Calgary Discretionary Advisor o CO •> "O < ± Event 1 a> CL "c 4 Event 2 O o • Event 3 cu CO CO i_ CD > < <N oo T - CD CD CD CD 1 - CO o O O O CNI CD o O O O CM CD o O O O CM CM T - O o o CM CO O o o T - CM CO O O O CM CD O O O CM CM T - O O CM CO O CM O O CM CO O CM O O CM Date (Months) There appears to be large increases in Average Net Contributions following Events 2, and 3. Table 3.5.6 shows the results from t-tests with equal variance assumptions. For each event, we tested the assumption that the Average Net Contributions one and a half years before the event was equal to the Average Net Contributions six months after the event. Table 3.5.6 summarizes our findings. Table 3.5.6 T-tests with Unequal Variance For Calgary Discretionary Advisors Variable Mean Net Contribution Standard Deviation Before Event 2 2,242,346 895,247 After E v e n t 2 4,002,191 1,550,000 Before Event 3 2,793,254 1,360,000 After E v e n t 3 3,175,829 1,780,000 Degrees of Freedom t - Value Pr > Itl 17 -2.08 0.0529 * 23 -0.38 0.7108 The results show that the change in Average Net Contributions following Event 2 may be statistically significant, as the P-value is slightly higher than 0.05. However, the change in Average Net Contributions per advisor following Event 3 was not proven to be statistically significant, as the P-value is shown to be greater than 0.05. 43 Unassigned Clients Figure 3.5.7 Unassigned Vancouver Clients t CO c o "5 A JO /S c o Ar /V o 8 o / A J 1 \ ^ # K \ V <\ A / yJt v\ / 4 B/ent 2 A n • B/ent 3 I m V I Oi o oo o> o> \ * A Event 1 00 oo o> O) o o> o> o> CD o o> o> 0> O) o ro O) O) CM o> O) O) CO o o o o CM CO o o o o CM OJ O O CM CM CM o o T - O o o OO O M - o o CM CD O o o CM OJ O CM O O CM O O CM ii- OO O CM O O CM CO O CM O O CM Date (Months) The Total Net Contributions for Vancouver clients show a sharp decrease during March 2000. Further investigation showed a series of clients withdrawing large sums at this time, which was not attributed to any particular event. We decided to remove this point and replace it with the average value of February and April 2000. The resulting data point is marked with a star. There does not appear to be large increases in Total Net Contributions of Vancouver clients 6 months after Events 1, 2, and 3. Table 3.5.7 T-tests of Publications Effect on Vancouver Client's Net Contributions Variable Before Event 1 After E v e n t 1 Before Event 2 Mean Net Contribution Standard Deviation 1,072,204 1,430,000 -431,088 3,490,000 443,755 2,190,000 After E v e n t 2 1,322,682 1,050,000 Before Event 3 1,645,529 1,600,000 After E v e n t 3 1,452,278 1,070,000 Degrees of Freedom t-Value Pr > Itl 5.79 0.66 0.5321 22 -0.61 0.5463 23 0.2 0.846 Our results show that none of the positive articles written about the firm and their funds had a significant impact on Net Contributions from Unassigned Vancouver clients. 44 Figure 3.5.8 Unassigned Toronto Clients The Total Net Contributions of Toronto clients do not show any large increases following Events 1, 2, and 3. Table 3.5.8 show statistical results from our hypothesis tests. Table 3.5.8 T-tests of Publications Effect on Toronto Client's Net Contributions Variable Before Event 1 Mean Net Contribution Standard Deviation 1,086,852 250,936 After E v e n t 1 628,147 473,516 Before Event 2 573,014 393,729 After Event 2 761,582 294,339 Before Event 3 563,151 367,876 After E v e n t 3 306,195 657,600 Degrees of Freedom t - Value Pr > Itl 17 1.83 0.0852 22 -0.7 0.4913 23 0.82 0.4223 The resultsfromour hypotheses show that, like the Vancouver Unassigned clients no significant changes in Net Contributions after each event can be seen for Toronto Unassigned clients. 45 Figure 3.5.9 Unassigned Calgary Clients 5.0 4.0 tn I 3 3.0 -2.0 CD O 00 CD CD CM T— 00 CD CD CM OT O OT CD OT OT OT OT OT OT OT OT CO CO o OJ OT o T— CO O O O O CM CO O O O O CM OT O O O O CM CM O O O CM CO O— CO O CM ^— OT O T o o CM o o CM O O CM O ^— o CM CO O CM O O CM CD O CM O O CM Date (Months) Calgary unassigned clients do not show an increase in total Net Contributions following Events 1,3, however, there does appear to be a slight increase in total Net Contributions following Event 2. Table 3.5.9 T-tests of Publications Effect on Calgary Client's Net Contributions Variable Before Event 1 After E v e n t 1 Before Event 2 Mean Net Contribution Standard Deviation 356,534 133,799 40,960 406,045 265,411 265346 1,356,660 478137 Before Event 3 620,098 428,879 After E v e n t 3 323,795 345,627 After E v e n t 2 D e g r e e s of Freedom t - Value Pr > Itl 17 1.1 0.2848 22 -3.03 0.0061 * 23 0.69 0.4941 The test for the effect of Event 2 on the Net contributions of Calgary clients yield a P-value of 0.0061, which is less than our critical value of 0.05. We, therefore, can reject our hypothesis that the Net Contributions of Calgary clients before and after Event 2 are the same. However, no effects on the Net Contributions of Calgary clients for Events 1 and 3 are found. 46 Figure 3.5.10 Unassigned Other Clients CD O 00 CD CD CNI T— CO CD CD CO O CD CD CD CD O CD CD CD CD O CD CD CD CNI CO O O O O CNI T— CD CD CD CO O O O O CNI CD O O O O CN CN CO O ^— o o o CD O CD O CN T O O CN CN O O CN O O CN — O o CN CO O CN O O CM CD O CM O O CN Date (Months) The above graphs do not show any changes in slopes of Cumulative Net Contribution before and after the four events for the unassigned clients. Therefore, we cannot conclude a relationship between positive press coverage and Cumulative Net Contribution. Table 3.5.3 shows the results from tests of changes in total Net Contributions following the three events. Table 3.5.10 T-tests of Publications Effect on Other Client's Net Contributions Variable Mean Net Contribution Standard Deviation Before Event 1 1,829,165 932,819 After E v e n t 1 1,038,464 909,517 979,331 569,128 After E v e n t 2 2,395,376 1,540,000 Before Event 3 1,492,499, 1,230,000 After E v e n t 3 1,247,903 Before Event 2 Degrees of Freedom t - Value Pr > Itl 17 1.12 0.277 22 -2.21 0.0378 23 0.32 0.7508 733,799 Our results showed, for all Other clients, not including Vancouver, Toronto, and Calgary customers, no increases in Net Contributions after Events 1 and 3. Following Event 2, however, we see a significant change in Net Contributions from Other clients. 3.6. Proximity Effect on Average Net Contributions To determine whether the new Calgary office, which was opened in November 1999, had an effect on Net Contributions from Calgary clients, we analyzed average Calgary Net Contributions for Vancouver and Calgary advisors. Figure 3.6.1 shows Average Net 47 Contributions per advisor, for both Calgary and Vancouver advisors. Thus, all contributions from Calgary clients, prior to November 1999, are made through a Vancouver advisor. Net ContributionsfromCalgary clients after November 1999, are either made through a Vancouver or Calgary advisor. Table 3.6.1 shows the resultsfroma t-test on the Average Net Contributions before and after the Calgary office. Figure 3.6.1 Average Net Contribution Per Non-Discretionary Advisor For Calgary Clients t o CO '> < ai 0- o CO - A CD O 00 O) O) 4u f v j V OO 00 O CD 0) o> CD Jl\ \ / \i \ A N. V V* i K A ¥- y cn OJ O o> 0 0 CM ) ) T - T - 00 O OJ o 0 0 ) ) T - 0 0 CM CO o o o o CM Date (Months) o> o o o o CM X o o o CM A T — *k CO O O) l 00 CO OJ CO o o o o o o o o o o CM CM N e w Office O p e n e d CM O O CM CM O O CM \ 1 i CO O CM O O CM .Estimated Mean From Figure 3.6.1, we can visually see a large jump in Net Contributions following the opening of the Calgary office. A t-test with equal variances was performed to determine statistical significance. The results are shown in Table 3.6.1. Table 3.6.1 T-test for Average Net Contributions Per Non-Discretionary Advisor New Office Mean Net Contribution Standard Deviation t - Value Pr > Itl -7.13 < .0001 0 - 39,771 120791 1 521,958 173973 Our results give a P-value that is less than 0.0001, which is statistically significant at the alpha level of 0.05. Therefore, we can conclude that the Average Net ContributionsfromCalgary clients before and after the opening of the Calgary office is not equal. A Non-Discretionary advisor is estimated to have lost an average of $39,771/month in Net ContributionsfromCalgary clients before the opening of the Calgary office. After the opening of the Calgary office, a NonDiscretionary advisor is estimated to have accumulated an average of $521,958/month in Net ContributionsfromCalgary clients. This is a 1412 % increase in Calgary Net Contributions. Residual analysis was conducted for the predictive model. Figure 3.6.2 show the variance in residuals for the predicted values. Figure 3.6.3 shows the normal probability plot of the residuals. 48 Figure 3.6.2 Residuals vs. Predicted values N o Office Office From this graph we see that the variance in residuals for both predicted values are quite different. The variance appears to have increased after a new office is opened, violating the constant variance assumption of normality. The mean values of the residuals are both close to 0 confirming one of the assumptions of normality. Figure 3.6.3 Normal Probability Plot of Residuals The normal probability plot of the residuals show that the residuals deviate slightly from Normal distribution. 49 Discretionary Advisors All Discretionary advisors from Vancouver and Calgary, who managed Calgary clients were analyzed to determine if on average, an advisor has experienced increased contributions from Calgary clientsfromthe period between October 1999 and May 2002. From the above graph we see that the variance in Net Contributions is increasing with time, thus a two-sample t-test with unequal variances was performed to test for differences in mean Net Contributions. Table 3.6.2 show the results of the t-test conducted with degrees of freedom calculated by the Satterthwaite approximation. Table 3.6.2 T-test for Average Net Contributions Per Discretionary Advisor New Office Mean Net Contribution Standard Deviation 0 201,554 290,651 1 870,432 508,709 t - Value Pr > Itl -3.64 0.0008 Table 3.6.2 shows a P-value of 0.008 is statistically significant at the alpha level of 0.05, thus we can conclude that the Average Net ContributionsfromCalgary clients after the opening of the Calgary office is not equal to the Net Contributions prior to opening the new office. An advisor acquired an Average Net Contribution of $201,554 per monthfromCalgary clients, compared to after the opening of the Calgary office, where an advisor is shown to have accumulated an average of $870,432 per month from Calgary clients. Thus after the opening of the Calgary office, there was an increase of 331% in Net ContributionsfromCalgary clients. 50 Figure 3.6.5 Residuals vs. Predicted Values 2000000 -500000 - 1000000 N o Office Office From this graph of the residuals, we see that the variance in the residuals is larger after the opening of a new local office, violating the constant variance assumption of the residuals. The mean value of the residuals is close to 0. Figure 3.6.6 Normal Probability Plot of Residuals hfarrml CXiant i I os From the normal probability plot, we see that the residuals slightly deviate from the normal distribution. Particularly at the tails of the distributions appear to deviate from the normal providing further evidence that the residuals may not follow a normal distribution. 51 3.7. Profitability Model and Scenario Analysis Tool In running several scenarios for the firm, we gained an understanding of the expansion plans generated from the scenario analysis tools. Management was particularly interested in how specific changes in the optimal recommendations affected their maximization criteria, internal rate of return . The specific changes in the optimal recommendations that were studied are outlined below. This thesis discusses some of the preliminary observations gained from the scenario analysis tool and relates these findings to the business practices of the firm. For reference, the parameters of the model are shown in Appendix B. We enlisted the help of the firm's management to determine a list of parameter changes that were the most relevant to their business practices. This included the occurrence of hiring one less Discretionary advisor, hiring one more Non-Discretionary advisor, a 2% increase or decrease to their projected market growth rate during the 10 year period, increasing or decreasing office costs by 50%, opening a local office in the foreign market, hiring additional advisors one year later than the optimal recommendations, and a 2.5% increase or decrease to the projected book growth rate, due to being in close proximity to their clients. By analyzing these changes, management would understand the variability in the expected profitability of each expansion plan. They would also understand the sensitivity of each variable on the expected profitability of each new expansion. We ran the model and optimized several scenarios for different locations as well as different maximum quarterly cash outlays willing to be invested by the firm. Figure 3.7.1 shows a typical sensitivity analysis report resulting from the maximized scenario run. Figure 3.7.1 Sensitivity Analysis On Maximized Internal Rate of Return Base Case Value Discretionary Advisor (-1) N o n - D i s c Advisor (+1) o •c co c cu o CO Market Growth (4%,8%) Office C o s t (+50%,-50%) Office Location (Existing Office) Start Date (1 Year Later) Proximity Growth Rate (0.025,0.075) IRR From our analysis, we observed that in general, it was more profitable to hire advisors earlier rather than later, despite the large investment of salary costs. The number of recommended Internal rate of return is used as the maximization criteria throughout this section although the scenario analysis tool can maximize over several other criteria. 3 52 advisors to hire is, thus, constrained on the maximum quarterly cash outlay the firm is willing to undertake. Our sensitivity report shows the effect of hiring advisors 1 year later than the suggested hiring plan found by our model. If this occurred, we the internal rate of return would decrease approximately 24%. Notice that this variable is causing the most change from the optimal base case. During all our scenario runs, this variable was consistently one of the most sensitive variables influencing the internal rate of return. Figure 3.7.1 also shows that the initial start-up cost of opening a new office is not very sensitive to the internal rate of return, as was initially hypothesized by management. Increasing the initial start-up costs by 50% only reduces the internal rate of return to 28.8%. This was invaluable to management who was reluctant to enter new markets as they felt the initial start-up costs would greatly impact their profitability. We also observe that hiring one more Non-Discretionary advisor to the optimal hiring plan did not reduce the IRR greatly at 4.1%. The model, generally would advise to hire more Discretionary advisors and, in many cases, no Non-Discretionary advisors. As the NonDiscretionary business is a vital component of thefirm,for servicing clients in close relation to their Discretionary clients but who do not meet the minimum requirement for Discretionary services, management wanted to understand how the hiring of a Non-Discretionary advisor to the optimal plan would impact their profitability. Therefore, despite the models recommendations, they would hire at least one Non-Discretionary advisor during the start-up of a new office. The results from the sensitivity report gave them a greater understanding of the impacts of their business decisions. To further their understanding of their expansion scenarios, we also analyzed the results of the sensitivity analysis on the amount of time the expansion plan would take to financially breakeven. Figure 3.7.2 shows the sensitivity report on the time to breakeven. Figure 3.7.2 Sensitivity Analysis On Time to Breakeven Base Case Value Discretionary Advisor (-1 N o n - D i s c Advisor (+1 Market Growth (8%,4% o c? Office C o s t (-50%,+50%) o w Office Location (Existing Office) Start Date (1 Y e a r Later) Proximity Growth Rate (0.025) 10 15 20 Number of Quarters 25 30 53 Similar results were obtained for the expansion statistic, Time to Breakeven. We found that hiring additional advisors 1 year later greatly affected the time for the investment to breakeven. In this case, we see an increase from 23 quarters to 28 quarters. All the other scenarios did not change the time to profitability by more than one year. We then wanted to analyze the quarterly amount of investment needed to implement the 10-year expansion plan. In particular, we wanted to determine how sensitive this figure would be to various changes to the original expansion plan. Figure 3.7.3 shows the sensitivity report of the maximum quarterly investment for an optimized expansion plan. Figure 3.7.3 Sensitivity Analysis On Maximum Quarterly Investment Base Case Value , _ 1 1 l , 1—I 1 Discretionary Advisor (-1) N o n - D i s c Advisor (+1) Market Growth (8%,4%) p | w Office C o s t (-50%,+50%) Office Location (Existing Office) Start Date (1 Y e a r Later) Proximity Growth Rate (0.075,0.025) 0 500 1,0H»estfb6M($)2,000 2,500 3,000 Figure 3.7.3 showed that the most influential variable to the maximum quarterly investment was the number of Discretionary and Non-Discretionary advisors hired. This result was interesting as it provided support for management's belief that advisor compensation was one of the largest determining factors for expansion planning. This result showed that the addition of a NonDiscretionary advisor or the subtraction of a Discretionary advisor to the new office changed the amount invested by approximately +$500,000, and -$500,000, respectively. The sensitivity analysis also showed that the initial start-up costs of a new office had a considerable impact on the maximum quarterly investment as an increase in initial office costs of 50% increased the maximum quarterly investment by $140,000. This increase was, however not as large as previously predicted by management. Thefirmhad previously modelled profitability and in general our analysis showed a slightly lower internal rate of return and a longer time to breakeven estimate compared to their results. This difference was a result of improved cost estimates in our model. 54 4. INTERPRETATION OF RESULTS Our results have given insight into management's questions concerning the effect of specific drivers on revenue growth. We have shown that both Discretionary and Non-Discretionary advisors who proactively market to new clients obtain increased book growth compared to their Reactive counterparts. We have seen 31.24% more growth for proactive Non-Discretionary advisors and 3.8% more growth for Proactive Discretionary advisors over a 5-year period. This information gives the firm the statistical evidence to support their hypothesis and the confidence to begin hiring Proactive advisors. Some of our statistical results regarding skill level affect on book growth contradicted managements beliefs. As management had categorized their advisors into High, Medium, and Low Skilled, they had assumed that our results would show that High Skilled advisors outperformed Medium Skilled advisors, while Medium Skilled advisor outperformed Low Skilled advisors. However, it was discovered that for Non-Discretionary services, advisors rated High Skilled showed no significant difference in performance compared to advisors rated Low Skilled. Whereas, advisors rated Medium Skilled outperformed both High and Low Skill rated advisors. For Discretionary services, our results showed that both High and Medium Skill rated advisors outperformed Low Skilled rated advisors. However, Medium Skill rated advisors showed increased growth rate compared to advisors rated High Skilled. Our results show that the difference in book growth between High and Medium Skilled advisors was not a large as originally perceived. As Medium Skilled advisors were thought to perform well most of the time but not always consistently, it is possible that advisors are not being classified correctly. Our results have given management confidence in their ability to assess advisor in terms of their marketing and in some ways their assessment of skill level. In hiring additional advisors to service new markets, they will aim to hire Proactive, High or Medium Skilled advisors. In our analysis of positive press coverage on thefirm,we have shown that there were significant increases for Vancouver and Toronto Non-Discretionary advisors and Calgary Discretionary advisors as well as for Calgary and 'Other' Unassigned clients following a favourable comment in a Nationally distributed newspaper on January 2001. We have also seen increases from Calgary Non-Discretionary advisors and Toronto Discretionary advisors following two articles in the November and December 2001 issues of a nationally distributed magazine. We cannot, however conclude that these events cause these increases but there does appear to be significant changes in Net Contributions for some advisors and Unassigned clients after these publications, which may or may not be attributed to thefinancialpress coverage. There is very substantial evidence that a local office in a targeted new market, impacts both the Discretionary and Non-Discretionary advisors accumulated client Net Contributions. A 1412% increases in Net Contributions is shown among Non-Discretionary advisors and a 331% increases in Net Contributions is shown among Discretionary advisors. However, it should be noted that we are comparing full time advisors who are placed in the local market to part time advisors outside of the target market. Therefore our results may be somewhat exaggerated. 55 Further analysis would be needed to compare a full time remote advisor to a full time local advisor, with respect the targeted market. Based on our preliminary results, the firm should strive to open local offices in their targeted markets as there will be a greater realization of revenue and, as our results have shown, start-up costs of a new office does not drastically affect profitability. The profitability model is currently being used by thefirmto develop expansion plans to convince shareholders to invest in additional offices. As they are still in the planning phases, no results from their implementation can be assessed to determine how closely our model predicts actual outcomes. However, our model is currently being used to assess the expected profitability of several previously opened offices to determine whether our model is accurately predicting their realized profits. 5. SUMMARY This paper discusses the findings from a collaborative study with the Centre for Operations Excellence at the University of British Columbia and a financial managementfirm.Thefirmis seeking to increase their market share of private investors in a number of Canadian cities. In order tb determine the most profitable method of expanding their business to service additional clients, we developed a scenario analysis tool to analyze operational expansion strategies in terms of profitability. This allowed management to determine an optimal expansion and hiring strategy to service additional markets while generating the largest expected profit for the firm. In addition to developing a profitability analysis tool, we analyzed and quantified the impact of several factors believed to impact revenue growth. These estimates allowed us to improve our forecasts of revenue growth and further customize the profitability analysis tool. 6. FUTURE DIRECTIONS AND COMMENTS A further extension of our analysis of factors influencing client net contributions would be to analyze the impact on the growth of the number of clients serviced by the firm. This would allow us to understand how these factors influence their network of clients. In our analysis of positive press coverage, a possible direction would be to considerfittingtimeseries models, with intervention analysis, to analyze the affect of positive press coverage on Average Net Contributions to the firm. This would allow us to not only see potential shift changes but also possible decaying effects throughout time. Another further extension to our analysis would be to incorporate seasonality effects in the Cumulative Net Contribution growth models. This could potentially improve forecasting of revenue growth by reducing variation around our forecasts. 56 One issue that the reader should be aware of is that all the data used in our analysis was collected during a bull market. Thus, our profitability model assumes that Net Contributions to the firm are unrelated to the stock market conditions. Secondly, our models for Book Value growth do not incorporate seasonality effects on growth. Lastly, our profitability model does not incorporate any adjustments for tax effect on profitability, thus taxation issues should be incorporated into any expansion analysis. 57 REFERENCES Berman O, Ganz Z, Wagner JM. A Stochastic Optimization Model for Planning Capacity Expansion in a Service Industry under Uncertain Demand. Naval Research Logistics, Vol. 41 (1994), pp. 545-564. Department of Finance Canada <www.fm.gc.ca/toce/2002/cmfi_e.html> Efroymason MA, Ray TL. A Branch-Bound Algorithm for Plan Location. Operations Research, Vol. 14 (1966), pp. 361-368. Franjic, Nicole. Theories of Investor Behaviour and Their Application to Segmentation and Predictive Modelling of Retail Clients at an Investment Management Firm. Vancouver: University of British Columbia. 2001. Geoffrion A, McBride R. Lagrangian Relaxation Applied to Capacity Facility Location Problems, AIIE Trans. Vol. 10 (1978), pp. 40-47. Larsen RJ, Marx ML. An Introduction to Mathematical Statistics and its Applications, 2nd edition, Prentice-Hall. Neter J, Kutner MH, Nachtsheim CJ, Wasserman W. Applied Linear Statistical Models. 4th ed. Irwin (1996). Ratkowsky, D.A. (1983). Nonlinear Regression Modeling: A Unified Practical Approach, Dekker, New York. SAS Online Documentation v8 <http://www.sfu.ca/sasdoc/sashtml/onldoc.htm> SAS Institute. SAS On-line Help Files. Shulman A. An Algorithm for Sloving Dynamic Capacity Plan Location Problems with Discrete Expansion Sizes. Operations Research, Vol. 39 (1991), pp. 423-436. Steel RGD, Torrie JH. (1980). Principles and Procedures of Statistics. 2nd ed. McGraw-Hill, New York. 58 APPENDIX A PROCESS MAPS: CLIENT ACQUISITION Non-Discretionary Advisors Referrals from existing clients & personal network Referrals from w i t h i n The Contact Centre Queue F i r m : Discretionary and Institutional C O I ' s Walk-in's 1 1 r. Contact prospect v i a phone call or email M a r k e t T h e F i r m to prospect No D o not pursue prospect further Yes Suggest other alternatives to prospect Determine what service levels the prospect is eligible to receive. (Non-Discretionary or Discretionary) • 59 Enter prospect into Client Management System and do not pursue further Yes Redirect prospect to Discretionary Management Prospect sends asset holding statements for review Advisor gathers prospects information: - Investment horizon (time) - Risk tolerance - Objectives - Tax issues Advisor creates asset mix strategy * Advisor recommends The Firms mutual funds to obtain asset mix strategy 60 No Yes Yes Send draft Investment Policy Document (1 or 2 weeks after meeting) 1T \r Phone Calls Meeting r Review Investment Policy Document and determine if changes are necessary Enter prospect into Client Management System and possibly pursues at a later date Yes Client is Unassigned Client is Assigned Review portfolio (Min 1/yr) Re-balance (If necessary) Retirement Planning ' (As necessary) Year-end Tax reporting Contact Centre (for servicing) Contact Advisor Queue (for advice) r Answer Calls Answer Emails 62 Discretionary Advisors Referrals from Existing Clients & personal network Referrals from within the firm Advisors Network of Professionals Seminars Contact Centre Walk-ins ~~T~ Contact prospect via phone call or email Enter prospect into client management system for possible future reference Add client to personal mailing list for the firm quarterly letter Pursue prospect via phone call or email Yes Take prospect out of personal mailing list No Is prospect still interested? Yes Yes Does prospect want to meet? 63 Meeting Phone Call Reroute prospect to Non-Discretionary department, or suggest other options Yes Suggest other alternatives to prospect Discuss prospects investment objectives r Send draft Investment Policy document (1 or 2 weeks after discussion of investment objectives) 1 64 Meetings Phone Calls Emails Portfolio Management Rebalancing (2 or 3 times/yr) Processing Transactions & reporting Contact * I Retirement Planning Year-end Tax reporting (1/yr) Meetings (1/yr) r Calls (1/qtr) Emails (1/yr) Frequency of contact varies across Portfolio Managers and depends on how long the client has been with the Portfolio Manager 65 APPENDIX B PROFIT MODEL PARAMETERS Advisor Specific Parameters Cost Item Direct Discretionary Advisor Cost (excluding travel costs) Discretionary Advisor Travel and Meals Cost Indirect Discretionary Advisor Cost Fixed Discretionary Cost (all offices) Bonus and Fees Discretionary Advisor Bonus on Cumulative Book Value (Quarterly) Discretionary Revenue / Client Fees on AUM (Quarterly) Start Dates Discretionary Advisor 1 Start Date (Enter "-" if no start date) Discretionary Advisor 2 Start Date (Enter "-" if no start date) Discretionary Advisor 3 Start Date (Enter "-" if no start date) Proximity % Increase in Discretionary Net Contribution Growth Due to the Presence of a Local Office % Increase in Non-Disc Net Contribution Growth Due to the Presence of a Local Office Proximity (0 = Service market through an existing office; 1= Open new office in market) Available Capacity in Existing Office (0 = No; 1= Yes) Market Potential Discretionary Market Potential Discretionary Market Share as Percentage Discretionary Market Share (Assets Under Management) Non-Discretionary Market Potential Non-Discretionary Market Share as Percentage Non-Discretionary Market Share (Assets Under Management) Inflation, Discount, and Growth Rates Inflation Rate for Cost (Quarterly) Discount Rate for NPV (Quarterly) Discretionary AUM Market Growth Rate (Annual, City Specific) Non-Discretionary AUM Market Growth Rate (Annual, City Specific) Office Costs Rent and Operating Costs Excl Insurance (Quarterly) Premise Insurance (Quarterly) Receptionist Salary and Benefits (Quarterly) Office Furniture, Signage, Leaseholder Improvement (One-time) Hiring and Business Planning Costs (One-time) Alarm/Internet Office Cost (Quarterly) 66
- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- Expansion planning at a prominent Canadian investment...
Open Collections
UBC Theses and Dissertations
Featured Collection
UBC Theses and Dissertations
Expansion planning at a prominent Canadian investment management firm Chung, Kelly Anne 2002
pdf
Page Metadata
Item Metadata
Title | Expansion planning at a prominent Canadian investment management firm |
Creator |
Chung, Kelly Anne |
Date Issued | 2002 |
Description | A prominent financial investment firm currently operates offices located in four major cities across Canada. In addition to their current markets, they are seeking to increase their market share of private investors in a number of Canadian cities over a 10-year planning horizon. Expanding their operations to service these new markets would mean either increasing the capacity of their current offices or acquiring additional offices located in new markets, both of which are costly. Thus, a critical decision must be made as to the most cost effective method of expanding their operations to service new markets. Though costs are expected to be greater for a new office location, it is believed that servicing clients from an office located within the prospective city will positively affect revenue growth for the firm. This assumption along with other factors believed to impact revenue growth, such as positive press coverage, are assessed to determine and quantify their potential impact. As advisors' compensation and support are a large component of the costs incurred to service new markets, a hiring strategy of advisors must also be developed for each new market entered. Results from both our scenario analysis tool as well as our analysis on factors affecting revenue growth have given management the information needed to convince shareholders to invest in expansion planning. Please note that all absolute numbers have been altered within this thesis to adhere to a confidentiality agreement with the financial management firm. |
Extent | 3804010 bytes |
Genre |
Thesis/Dissertation |
Type |
Text |
File Format | application/pdf |
Language | eng |
Date Available | 2009-10-19 |
Provider | Vancouver : University of British Columbia Library |
Rights | For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use. |
DOI | 10.14288/1.0090913 |
URI | http://hdl.handle.net/2429/14059 |
Degree |
Master of Science - MSc |
Program |
Business Administration |
Affiliation |
Business, Sauder School of |
Degree Grantor | University of British Columbia |
Graduation Date | 2002-11 |
Campus |
UBCV |
Scholarly Level | Graduate |
Aggregated Source Repository | DSpace |
Download
- Media
- 831-ubc_2003-0207.pdf [ 3.63MB ]
- Metadata
- JSON: 831-1.0090913.json
- JSON-LD: 831-1.0090913-ld.json
- RDF/XML (Pretty): 831-1.0090913-rdf.xml
- RDF/JSON: 831-1.0090913-rdf.json
- Turtle: 831-1.0090913-turtle.txt
- N-Triples: 831-1.0090913-rdf-ntriples.txt
- Original Record: 831-1.0090913-source.json
- Full Text
- 831-1.0090913-fulltext.txt
- Citation
- 831-1.0090913.ris
Full Text
Cite
Citation Scheme:
Usage Statistics
Share
Embed
Customize your widget with the following options, then copy and paste the code below into the HTML
of your page to embed this item in your website.
<div id="ubcOpenCollectionsWidgetDisplay">
<script id="ubcOpenCollectionsWidget"
src="{[{embed.src}]}"
data-item="{[{embed.item}]}"
data-collection="{[{embed.collection}]}"
data-metadata="{[{embed.showMetadata}]}"
data-width="{[{embed.width}]}"
async >
</script>
</div>
Our image viewer uses the IIIF 2.0 standard.
To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0090913/manifest