Analyzing the Eﬀects of Monetizing Intangibles in Decision Analysis A Case Study for Video-Conferencing

(1)

Analyzing the Effects of Monetizing Intangibles in Decision Analysis

A Case Study for Video-Conferencing

B.R. Mateman s1536257

Den Haag, April 29, 2011

Master Thesis Econometrics, Operations Research and Actuarial Studies Specialization: Operations Research

Supervisors:

(2)

(3)

Analyzing the Effects of Monetizing Intangibles in Decision Analysis

A Case Study for Video-Conferencing

Master Thesis Econometrics, Operations Research and Actuarial Studies Den Haag, April 29, 2011

Author: Bart Mateman BSc

1536257

bartmateman@msn.com

Student Econometrics, Operations Research and Actuarial Studies Specialization Operations Research

University of Groningen

Supervisors: Dr. J.W. Nieuwenhuis (University of Groningen)

Ir. C.T.A de Vos (TNO)

Co-assessor: Prof. Dr. R.J.M. Alessie

Abstract: This thesis describes the monetization of qualitative aspects and their

incorpo-ration in decision analysis. In the context of corporate social responsibility, eco-nomic, environmental and social costs are becoming more important in decision making. To make a well-founded decision, it may help to quantify and monetize the qualitative aspects.

To explore the effects of taking into account monetized values of intangibles, a case study is performed. This case study consists of an investment decision prob-lem for video-conferencing and a decision probprob-lem for giving stimuli to reallocate meetings. The case study is described by an integer non-linear problem. A con-joint study is used to estimate the allocation of meetings given a set of stimuli. By a willingness-to-accept type of argument, the added value of the availability of video-conferencing systems, as perceived by employees, is determined. The ana-lytic hierarchy process is used to measure the relative importance of the economic, environmental and social costs in the model. Analyzing the results, it turns out that incorporating monetized social and environmental aspects strengthens the argument to invest. Taking into account monetized values of social aspects has more impact, than taking into account environmental costs.

Keywords: Decision analysis, corporate social responsibility, video-conferencing, conjoint

(4)

(5)

Management summary

Corporate social responsibility is becoming more important in decision analysis. In this context, when evaluating an investment decision, firms are interested in the consequences for the firm, environment and society. How to determine the effects of an investment? Not every effect will be trivial. Some will be qualitative while others will be of a quantitative nature. To make a well-founded decision, it may help to quantify the qualitative aspects. When the most relevant effects are quantified, trade-offs can be made between the results for the various objectives of the firm. Finally the firm can make a decision, such that it maximizes its CSR objectives.

However, quantification may not always be satisfactory. Some firms may be interested in results of decisions, expressed in monetary terms. It may help the decision maker to quantify and mone-tize effects, before making value judgements. This research describes how qualitative terms can be monetized, such that they can be taken into account in decision making for a particular case study.

The research was conducted at the department of Strategic Business Analysis of TNO in Delft. TNO is an independent research organization in The Netherlands. The department of Strategic Business Analysis aspires to develop expertise on the valuation of intangibles, such that intangi-bles can be taken into account in business models. In the context of this aspiration a research is conducted with the objective:

Analyze the effects of incorporating monetized values of qualitative terms in decision analysis.

For this purpose a case study is investigated. An investment decision problem for video-conferencing is considered. Besides making this investment decision, the firm can decide to stimulate employees to change their allocation of meetings. On the basis of this case study, the effects of incorporating monetized values of qualitative terms in decision analysis are evaluated.

(6)

of employees, a conjoint study is performed. As a result, given a set of stimuli, the allocation of meetings over types of meetings can be predicted for meetings at three indications of distance. Furthermore, the data from the conjoint study is used to estimate the added value of the availabil-ity of video-conferencing systems as perceived by employees. Environmental costs are calculated by a life cycle assessment procedure developed by the Institute of Environmental Sciences in Lei-den. The analytic hierarchy process is used to measure the relative importance of the economic, environmental and social costs in the model.

From the conjoint study we find that meetings at nearby locations can be influenced more easily by stimuli, in the form of vacation hours, than meetings at distant locations. A lower bound for the social value added of video-conferencing is 365 vacation hours per employee per year, that is e6,146. As a result, the social value created by having video-conferencing systems available is higher than the investment costs are. Taking into account environmental costs has less impact, since the environmental costs are small in comparison with the economic and social costs. In any case the firm is advised to invest in video-conferencing systems.

Given that a firm wants to stimulate employees by means of giving vacation hours per meeting, minimizing the broad CSR definition of costs results in a different optimal solution than when minimizing economic costs. Using the broad definition results in stimulating more than when the conventional economic definition is used. This is due to the fact that giving stimuli is cheaper when the given compensations are perceived as social value added. In both solutions tele- and video-conferences are stimulated. Both solutions are sensitive to the specification of parameters in the model. Changing a parameter often results in a different optimal solution. This is due to the fact that several solutions give near optimal values. It could be that the model overestimates the use of video-conferencing. In that case it is not clear whether the company should stimulate the use of particular types of meetings.

(7)

Preface

This thesis is the final project for my study Econometrics, Operations Research and Actuarial Studies at the University of Groningen. It is the result of an internship at the department Strate-gic Business Analysis of the Dutch independent research organization TNO in Delft. I am grateful that TNO gave me the possibility to do my final project as an intern. I experienced it as a valuable and nice time.

I would like to use this preface to thank the people who helped me during my final project. Espe-cially, I would like to thank my supervisors J.W. Nieuwenhuis, from the University of Groningen and Coen de Vos, from TNO, for their critical and useful comments and support. Furthermore, I would like to thank Frank Berkers, Maarten Hoeve, Hugo Gelevert and other colleagues for their help and input. Finally I wish the reader pleasure reading this thesis.

(8)

(9)

Introduction

This chapter gives an introduction of the thesis. First the subject of this study is introduced. Then the outline of the thesis is presented and the case study is introduced. This chapter concludes with an introduction of TNO, the company where this research was conducted.

1.1 Incorporating qualitative aspects in decision making

Decision analysis deals with decision problems too complex to solve by simple common sense. The complexity of decision problems is caused by several factors including multiple objectives, intan-gibles and value trade-offs. These factors have a special role in evaluating decision problems in the context of corporate social responsibility. Corporate social responsibility (CSR) has its roots in the late 1960s and early 1970s. It is a form of corporate self-regulation where, besides economic prosperity, social and environmental prosperity are incorporated in the business model. More and more firms are concerned with doing business in a more environmental friendly and social way. They try to find a balance between people, planet and profit, i.e. the so-called triple bottom line.

In this context, when evaluating an investment decision, firms are interested in the consequences for the firm, environment and society. How to determine the effects of an investment? Not every effect will be trivial. Some will be qualitative while others will be of a quantitative nature. To make a well-founded decision, it may help to quantify the qualitative aspects. When the most relevant effects are quantified, trade-offs can be made between the results for the various objectives of the firm. Finally the firm can make a decision, such that it maximizes its CSR objectives.

Quantification is however not always satisfactory. Some firms are interested in results of decisions, expressed in monetary terms. Effects have to be quantified and monetized before making value judgements. No general method for monetization of qualitative terms is available, i.e. for each specific problem a particular method has to be selected or developed to monetize intangibles.

Figure 1.1 shows in blue how decision problems are treated in this thesis. We start with a decision problem. Relevant qualitative and quantitative aspects are monetized and value judgements are made between the various monetized aspects. Subsequently, the decision problem can be solved. An extension to this scheme is the determination of cashability of the monetized aspects. Casha-bility is the rate at which the monetized terms can be incorporated in the firms balance sheet.

For example, when a firm reduces environmental costs withe10,000 and the cashability rate is 50

percent, then the costs on the balance sheet are decreased bye5,000, or equivalently the profit

(14)

scope of this research.

Figure 1.1: Scheme of decision process in this thesis

This thesis describes how qualitative terms can be monetized such that they can be taken into account in decision making. The effects of incorporating monetized values of qualitative aspects in decision making are investigated. The situation where only economic terms are considered is compared with the situation where economic, environmental and social aspects are taken into account. A specific case study for video-conferencing and the allocation of meetings is performed. On the basis of this case study the strengths, weaknesses and the added value of monetization in decision analysis are discussed.

1.2 Outline of thesis

The outline of this thesis is as follows. In Chapter 2 the problem is formulated. In this chapter the context and objective of research, the research questions and the methodology to answer the research questions are presented. In Chapter 3 the mathematical model is derived. First the model is described, then a general mathematical model is formulated and finally the specific mathematical model is determined. Chapter 4 describes how conjoint analysis is used to model choice behavior. In this chapter the statistical foundations, the estimation procedure, experimental design and results are presented. Conjoint analysis is also used to monetize the added value of the availability of video-conferencing systems as perceived by employees. Chapter 5 discusses how priority weights for various objectives can be determined by making pairwise comparisons. For this purpose the analytic hierarchy process is used. The chapter describes the mathematical foundation, limitations and the results. In Chapter 6 other input parameters for the model are estimated. Economic, environmental and social costs are determined and used as input for the mathematical model. Chapter 7 describes the method used to solve the mathematical model and presents the results from solving the model. Subsequently, sensitivity analysis is performed. The main findings and recommendations for further research can be found in Chapter 8. The Glossary explains the abbreviations used in the text. Appendix A up to F present figures, tables and other not directly relevant information for the main text.

1.3 Case study

(15)

Introduction

for video-conferencing is considered. Should a company invest in video-conferencing systems or not? After making the investment decision, the decision maker can decide to give stimuli to reallocate meetings. Meetings can be allocated over face-to-face meetings, tele-conferences and video-conferences. We investigate whether a company should stimulate one or more types of meetings. Economic, environmental and social aspects are monetized and incorporated in the mathematical model, so that the effects of using a broader definition of costs can be investigated.

1.4 TNO

(16)

(17)

Chapter 2

Problem formulation

This research is done in the context of corporate social responsibility and decision analysis. Corpo-rate social responsibility is becoming more important in business, and therefore its role in making (investment) decisions is also increasing. Making decisions may be a difficult task, since the effects of alternatives on environment, safety, job satisfaction, etc. are not always clear. Those aspects are of a qualitative nature and difficult to measure. This thesis describes how qualitative terms can be measured and valued such that they can be taken into account in decision making. For this purpose a case study for video-conferencing is evaluated. Before stating the research objective and questions, the concepts of corporate social responsibility and decision analysis are introduced. Subsequently the case study, methodology and level and scope of the research are discussed. The problem formulation is finished by a literature review on the quality of meetings.

2.1 Context of research

This section describes the context of research. Corporate social responsibility and the difficulties regarding its definition are discussed. Subsequently decision analysis and the complexity of decision problems are discussed.

2.1.1 Corporate social responsibility

(18)

Elkington (1997), where the author argues that economic prosperity, environmental quality and social justice are performance measures of sustainability. Other dimensions covered in definitions of CSR are stakeholders (88 percent) and voluntarisms (80 percent) (Dahlsrud, 2008). Economic, environmental and social dimensions are covered in respectively 86, 59 and 88 percent of the def-initions of sustainability.

In this thesis, the definition of CSR of the European Commission is used. The European Commis-sion defines corporate social responsibility as “a concept whereby companies integrate social and environmental concerns in their business operations and in their interaction with their stakeholders on a voluntary basis”(European Commission, 2001). After this basic introduction on sustainabil-ity we continue with a concise introduction in decision analysis.

2.1.2 Decision analysis

Decision analysis is defined as ‘a philosophy, articulated by a set of logical axioms, and a method-ology and collection of systematic procedures, based upon those axioms, for responsibly analysing the complexities inherent in decision problems’ Keeney (1982). More intuitively decision analysis is a formal way to tackle decision problems that are too complex for the application of simple common sense. Decision analysis is founded on a set of axioms stated in Von Neumann and Morgenstern (1947). These axioms describe how to behave rationally in a social economy. The solution is a set of rules which tells the decision maker how to behave in every situation. The desirability (or total utility) of an alternative depends on the likelihood of its effects and on the preferences of the decision maker for those effects. As a result of the axioms, expected utility levels can be calculated for each alternative. The alternative with the highest expected utility is the most preferred alternative.

Decision analysis is used because of the complexity of many decision problems. Several factors in-fluence the complexity of decision problems, e.g. multiple objectives, difficulty of identifying good alternatives, intangibles, long-time horizons, many impacted groups, risk and uncertainty, risk to life and limb, value trade-offs and risk attitudes. For a more detailed overview and explanation of the complexities in decision analysis, the reader is referred to Keeney (1982).

2.2 Research objective and research questions

This thesis focuses on three of the aspects discussed in Section 2.1.2. We consider decision problems in the context of CSR. That is, we consider problems with multiple objectives, where intangibles and value trade-offs are important. If a firm takes into account CSR when making decisions, the decision maker may have several conflicting objectives. For example, a firm wants to minimize costs, deliver maximum quality, minimize environmental damage and maximize employee satis-faction. Minimizing costs may result in higher environmental damage or less satisfied employees. If multiple objectives are evaluated, trade-offs have to be made. What is more important to the

firm, cost reduction of ten percent or a reduction of CO2emissions by five percent? And how does

(19)

Problem formulation

and the effects of taking into account these monetized intangibles are unclear. This leads to the objective of this research:

Analyze the effects of incorporating monetized values of qualitative terms in decision analysis.

For this purpose an investment decision problem for video-conferencing is considered. Besides making this investment decision, the firm can decide to stimulate employees to change their allo-cation of meetings. On the basis of this case study, the effects of incorporating monetized values of qualitative terms are evaluated. Research questions to be answered are:

• How can a mathematical model be formulated that describes the decision problem?

• How can valuations for the tangible and intangible objectives be found?

• How can the value trade-offs made by the decision maker be quantified?

• How can the decision problem be solved?

• What is the effect of stimuli on the allocation and costs of meetings?

• What is the effect of the incorporation of intangibles on the decision made?

2.2.1 Introduction to the case study

We consider an investment decision for video-conferencing systems for TNO at the locations Delft, Groningen and Enschede. TNO has already invested in video-conferencing systems, but this is not relevant for the research. Important factors, when making an investment decision for video-conferencing systems in the context of CSR are environment, employee preferences and costs. Video-conferencing may save lots of money and harms the environment less then face-to-face

meetings at a distant locations. However, the usage of a system depends on the preferences

of employees. When employees decide not to make use of video-conferencing no advantages are realized. A decision maker can decide to stimulate employees to have more meetings of a particular type, such that the firm’s objectives are optimized. Before discussing these decisions, meetings are discussed in more detail.

2.2.2 Meetings and perceptions on meetings

A meeting is defined as ‘a gathering of two or more people that has been convened for the purpose of achieving a common goal through verbal interaction, such as sharing information or reaching

agreement.’1 _{In this thesis four types of meetings are considered, which are defined below:}

• Video-conferencing. A set of interactive telecommunication technologies which allow two or

more locations to interact via two-way video and audio transmissions simultaneously. 2

• Tele-conferencing. A set of interactive telecommunication technologies which allow two or more locations to interact via two-way audio transmissions.

• Physical (face-to-face) meeting, traveling by car. A meeting where participants have direct face-to-face contact and at least one of the participants travels by car to reach the meeting location.

1_{http://en.wikipedia.org/wiki/Meeting}

(20)

• Physical (face-to-face) meeting, traveling by public transport. A meeting where participants have direct face-to-face contact and at least one of the participants travels by public transport to reach the meeting location.

In the Netherlands employees participate on average in 2.2 meetings a week. This results in 3 hours and 18 minutes of meeting time a week (Vennik and Gabreels, 2007). At TNO, work is mainly project based and TNO does a lot of consulting work. Therefore, we expect that employees of TNO participate in more meetings than an average employee in the Netherlands. Meetings are not always at the office and therefore employees have to travel. Average traveling time to meetings is 1 hour and 36 minutes a week. How this relates to the average traveling time to meetings of employees of TNO is unclear. One can imagine that employees do not like to travel, since traveling could be considered as a waste of time. Furthermore, traveling harms the environment

due to emission of carbon dioxide. Also employers are harmed when their employees are on

the road, since traveling time is not always used efficiently. Note that business, employees and environment can be affected by meetings. These three aspects are covered by the triple bottom line of sustainability. It can be argued that having a meeting in Groningen, while your office is located in Delft, is not sustainable. A more sustainable alternative may be video-conferencing or tele-conferencing, where the environment is harmed less, employees may be more satisfied and costs may be reduced. We continue by discussing the perceptions of employees, employers and suppliers of video-conferencing solutions.

Perceptions of employees

In the Netherlands 44 percent of business travelers make use of video-conferencing (Jongsma, 2010). The frequency of video-calls is not mentioned in the report. Not all meetings are suitable for this type of communication. Jongsma (2010) reports that employees prefer to have personal meetings in the case of job interviews, sales talks and job evaluations. Furthermore, employees seem to prefer a personal meeting once in a while, also when video-conferencing or tele-conferencing is suitable. Barriers to make use of video-conferencing are the availability of video-conferencing systems (52.2%), the preference of face-to-face contact (52.2%), the perception of being less effec-tive using video-conferences (12.2%) and the perception that meeting partners prefer face-to-face meetings (24.3%).

Perceptions of employers

A report by Cisco, a supplier of information communication technology solution, gives reasons why companies use or do not use video-conferencing solutions. Users perceive the following advantages of video-conferencing: reduction of business travel, improvement of long distance communication, savings of money, increase of work life balance, enhancement of environmental responsibility, in-crease of productivity, helps maintaining business continuity, improvement of group collaboration and competitive advantage. The companies that do not use video-conferences give as main reasons lack of experience, high installation costs, lack of budget and perceived as unnatural (Cisco, 2010).

Suppliers of video-conferencing solutions

Publications of providers of video-conferencing solutions typically present major advantages and cost reductions due to video-conferencing. For example TANDBERG, one of the leading providers of video-conference equipment, reports in monetary terms on traveling cost reductions and

produc-tivity gains. Environmental impact reductions are reported in terms of CO2 emission reductions.

(21)

Problem formulation

assumes that 30 percent or more of the business travels are replaced by video-conferencing, when invested in video-conferencing systems. This percentage is not accompanied by data, but is based on experience of the company.

2.2.3 Description of objectives in the case study

So far we have discussed the perceptions on video-conferencing and the issues of meetings. We continue by discussing the decision problem a firm may face, when considering to invest in video-conferencing systems or when considering to stimulate employees to reallocate meetings. The firm makes a decision such that it optimizes its objectives. Possible objectives for the firm are min-imizing costs, environmental damage and risk and maxmin-imizing efficiency of meetings, employee satisfaction and customer satisfaction. Some of these objectives are of quantitative nature, while others are more qualitative. Note that the conventional approach minimizes costs only.

Economic costs of meetings consist of traveling costs, non-productive traveling time and accommo-dation costs. Accommoaccommo-dation costs occur when a hotel is booked or when dinner and/or lunch are consumed and paid for by the firm. Environmental costs are due to energy usage of equipment as notebooks, cell phones, etcetera. Also emission caused by traveling contributes to environmental costs. Minimizing risk in this context means minimizing the risk of being late due to traffic conges-tions, and minimizing the risk of having an accident, or minimizing the risk of technical failures. The efficiency and quality of a meeting may depend on the duration of the meeting, quality of communication, the quality of decisions made and the degree of participation in the group. Some participants may be more reluctant to speak freely during a meeting than other participants. Em-ployee satisfaction can be influenced by the quality of meetings, the time consumption of meetings and the possibility of having ad-hoc meetings. When meetings can be arranged on short notice, knowledge can be accessed or spread faster. Punctuality, knowledge accessibility and quality of communication are factors that can influence customer satisfaction in the context of meetings.

2.3 Methodology

This section reports on the methodology used to answer the research questions stated in Section 2.2. First a mathematical model is be formulated, then the aspects of the various objectives are monetized, subsequently value trade-offs are made between objectives and finally the model is solved.

2.3.1 Formulating a mathematical model

Before formulating the mathematical model for the case study, the model’s objectives, constraints,

input and output are described. Subsequently, a general model is formulated. This general

mathematical model is adapted such that it represents the model of the case study. We are dealing with an integer decision problem, as the investment decision is discrete and the compensations are also modeled in discrete terms. The mathematical model is formulated in Chapter 3.

2.3.2 Valuation of objectives

(22)

traveling costs, operational costs of meetings, traveling time and the productivity of employees during traveling are required. These data can be gathered relatively easy and enable us to calculate the economic costs of the various types of meetings. To evaluate environmental impact, a life-cycle-assessment method developed by the Institute of Environmental Sciences in Leiden is used. This method determines the impact of usage, production, transport and disposal of products to the environment. This impact is converted to monetary terms by a shadow price method. The shadow price represents the required compensation to restore environmental damage.

2.3.3 Value trade-offs between objectives

This study tries to monetize objectives of a decision problem such that a decision can be made with the least cost (or the highest value). However, it is questionable whether these monetized values are actually collectable for the decision maker. Does the decision maker have to pay for the environmental costs he causes with his operations? Will the value employees attach to the availability of video-conferencing be cashed when a company invest in video-conferencing systems? The decision maker may consider one objective more important than another. A manager has to evaluate the three objectives and weight them. Does he consider the objectives as equally important or not? In this research, the decision maker is interviewed, to find out his preferences for the case study.

2.3.4 Solving the decision problem

Integer decision problems can be solved by complete enumeration, or by making use of algorithms and heuristics. An optimization method is selected when the complete structure of the model is known. The method to be chosen depends on the size, linearity and convexity of the problem. The model is solved when intangibles are taken into account and when not. Sensitivity analysis is performed to evaluate the sensitivity of the solutions to the parameters. The results are discussed in Chapter 7, where the effects of incorporating intangibles are also presented.

2.4 Level and scope of research

To answer the research questions we limit the level and scope of the research. As discussed

before we consider a case study. We limit ourselves to a model with three objectives. Employee valuation, operational and environmental costs are taken into account in the objective function. A more advanced study could also incorporate employer attractiveness, quality of decisions in meetings, social costs of traffic jams, knowledge accessibility, work live balance of employees, potential housing costs, efficiency of meetings, etcetera. Besides the limitation on the objectives we limit the dynamics of projects. We consider a world where projects can be at three locations, i.e. Den Haag, Utrecht and Groningen. The office of the firm is located in Delft and it is assumed that for each project one employee of the location in Delft is participating in meetings. Furthermore the length of a project is fixed and given (ten meetings of variable duration per project). We assume that each year employees have a fixed number of meetings in the three given cities. Tele-conferencing is assumed to be available for all employees of the firm. Furthermore, we consider a firm where employees can decide for themselves how to have meetings.

2.5 Literature review on quality of meetings

(23)

Problem formulation

section literature on the quality of computer mediated communication and face-to-face meetings is reviewed. We come to the conclusion that there are reasons to believe that the quality of face-to-face meetings is comparable with the quality of tele- and video-conferences.

In Kane and Luz (a) the effectiveness of video-conferencing in patient case discussions is investi-gated. Kane and Luz (a) compares multi-disciplinary medical team meetings in a video-conference and co-located setting. Medically trained observer participants are judged on their ability to cap-ture information and making right decisions in the two settings. The results suggest that partici-pants are more eager to give their opinion in video-conferences. Furthermore, respondents believe that they better follow the discussions and understand the reasoning behind decisions better in video-conferences then in co-located meetings. However, co-located meetings seem to have fewer errors and more critical evaluations, while in video-conferences participants do not express their feelings of disagreement with patient management decisions. In an earlier report, Kane and Luz (b) finds that in multi-disciplinary medical team meetings participants have the need to see all partici-pants in the discussion. There is also the need to see the facts and data on patients to be discussed.

Careau, Vincent, and Noreau (2008) discusses the effectiveness of video-conferencing for a reha-bilitation team, developing an inter-professional care plan. The results of the study should be interpreted with care, since a sample of thirteen video-conferences is used. Careau et al. (2008) re-ports a productivity level of 96 percent during the meetings. The average length of the meetings is sixty minutes and comparable to face-to-face meetings. It is suggested to have the video-conference coordinated by a person who ensures a sound structure and gives every participant the right to speak. Interferences are considered as useful, but the comments are difficult to hear. The poor audio quality is the main disadvantage of video-conferencing. The good visual contact is consid-ered as the main advantage of video-conferencing.

Boyle, Anderson, and Newlands (1994) finds that performing a problem solving task is easier when there is visual contact. Less words and time are needed when using video-conferences instead of tele-conferences. However, O’Mally, Langton, Anderson, Doherty-Sneddon, and Bruce (1996) finds a contrary result. Cases where video communication is used results in longer and more interrupted dialogues. The performance of tele-conferences and video-conferences are the same. Comparing video-conferencing and face-to-face meetings provides similar results, where face-to-face meetings have shorter and less-interrupted dialogues. A later study finds similar results, see O’Mally, Lang-ton, Anderson, Doherty-Sneddon, Bruce, and Garrod (1997).

Bos, Olson, Gergle, Olson, and Wright (2002) reports on the establishment of trust at a distance. Four different types of communication are considered: face-to-face, video, audio and text chat. The research shows that text chat is not a sound base for trust. Video and audio conferencing are significant improvements but both suffer for delayed trust and fragile trust. It takes more time in those settings to gain trust and cooperation than in face-to-face meetings. Face-to-face meetings have the best base for a trustworthy relation.

(24)

(25)

Chapter 3

Mathematical model

This chapter describes how the mathematical model is built. First input, output, objectives and constraints are discussed. Subsequently the relations in the model are determined. Then a general mathematical model is formulated, which is specified for our particular case study.

3.1 Model description

This section describes the model of the decision problem. Employees of a firm have to participate in many meetings a year. Meetings can be held in several ways, i.e. employees can take part in a physical meeting, a tele-conference or a video-conference. Traveling to physical meetings can be done by car or by public transport. A firm wants to minimize the total costs of meetings. The total costs of meetings can be described by the economic, social and environmental costs of meetings. To minimize total costs the firm can make video-conferencing available and stimulate employees to participate in meetings with lower costs. The remainder of this section describes the input, output, the objectives and the constraints of the mathematical model.

3.1.1 Input and output

The input of the model consists of the following data:

• Costs. Costs are important in the decision problem. We consider operational and fixed costs per year for economic, environmental and social aspects of meetings. The fixed costs are given by the investment costs. The operational costs consist of traveling costs, costs of non-chargeable hours of meetings, costs of technical equipment, stimulation costs (if applicable) and environmental costs. If stimuli are given social costs are reduced. Furthermore, employ-ees attach value to the availability of a video-conferencing system. This value is measured per meeting and is also denoted as social benefits. Traveling costs are the compensations for traveling-expenses. Hours are non-chargeable when employees are waiting or when em-ployees are traveling to a meeting without using their traveling time for business purposes. Compensation costs are paid to employees when the firm decides to stimulate one or more

types of meetings. Environmental costs are the costs of CO2 emission caused by a meeting.

Costs of technical equipment are the costs for usage of technical equipment like bandwidth and setting up a connection.

(26)

of the decision maker for economic, environmental and social costs have to be determined. In case cashability of economic, environmental en social aspects is known, the weights could represent the rate of cashability. The determination of cashability is however out of the scope of this research.

• Employee behavior. To find the total costs of meetings, information on employee behavior is required. How do employees allocate meetings over the four different types of meetings at specific locations? And how do employees change their behavior when a specific type of meeting is stimulated at a location? How often and at what location do employees have meetings? These data are required to find the expected allocation of meetings per location.

Given the total number of meetings a year at the various locations, the stimuli and the investment decision, the model gives the number of meetings for each type of meeting per location. Given this allocation of meetings, the economic, social and environmental costs can be determined. Taking into account the relative importance of the various cost components, the total costs of meetings can be calculated.

3.1.2 Objective and constraints

The objective of the firm is to minimize total costs of meetings, by setting stimuli and making an investment decision for video-conferencing. When investing in a video-conferencing system, it is assumed that the firm invests in such a way that demand for video-conferencing can be met. Stimuli can take on only non-negative discrete values and only one stimulus can be chosen for a type of meeting at a location.

3.1.3 Graphical representation of the relations in the model

In Appendix A, a graphical representation of the relations in the model is presented. Invest-ing in video-conferencInvest-ing systems increases the fraction of video-conferences, increases economic costs, increases environmental costs and decreases social costs. Investment costs are the economic costs. Environmental costs are not increased since energy and resources required for production of video-conferencing systems are taken into account in the operational environmental costs of video-conferencing. Social costs are decreased, since the investment creates value for employees. As the fraction of video-conferences increases, the fraction of the other three types of meetings should decrease.

Stimulating a particular type of meeting increases economic costs, decreases social costs, increases the fraction of that type of meeting and decreases the fraction of all other types of meetings. The increase in economic costs is caused by the costs of stimulation and social costs are decreased, since employees receive extra compensation.

An increase of the fraction of meetings of a particular type results in an increase in economic and environmental costs. These costs represent the costs of having meetings of a particular type.

3.2 Model specification

(27)

Mathematical model

3.2.1 General model

The following index sets are defined:

i ∈ {1, . . . , I} Respondent index.

j ∈ {1, . . . , J } Type of meeting index.

k ∈ {1, . . . , K} Location index.

l ∈ {1, . . . , L} Compensation level index.

m ∈ {1, . . . , M } Cost component index.

We define the following parameters:

I Number of respondents.

J Number of types of meetings.

K Number of locations.

L Number of compensation levels.

M Number of weights.

Nik Number of meetings for respondent i at location k.

Nk Number of meetings at location k per year.

µik Weight attached to respondent i for meetings at location k.

wm Weight attached to cost component m.

OCjkm Operational costs for cost component m for meeting of type j at location k.

F Cm Yearly fixed costs for cost component m when invested in video-conferencing

system.

We define the following unknown functions depending on the decision variables:

fijk(X) Fraction of meetings of type j at location k for respondent i

given the vector with decision variables X = {x1kl, . . . , xJ kl},

xjkl is a 1 × L vector ∀j ∈ {1, . . . , J }, ∀k ∈ {1, . . . , K},

and video-conferencing is included in the set of meetings types. ˆ

fijk(X) Fraction of meetings of type j at location k for respondent i

given decision variable X and video-conferencing is not included in the set of meeting types.

fijk(X) and ˆfijk(X) have the property that for fixed X,

J X j=1 fijk(X) = J X j=1 ˆ

fijk(X) = 1, ∀i ∈ {1, . . . , I}, ∀k ∈ {1, . . . , K}.

Furthermore, we have that fijk(X) ≥ 0 and ˆfijk(X) ≥ 0, ∀i ∈ {1, . . . , I}, ∀j ∈ {1, . . . , J },

∀k ∈ {1, . . . , K}. The first property states that the fractions of all meeting alternatives must sum to unity for both the case when video-conferencing is available as the case when video-conferencing is not available. The second property states that the fraction of a meeting alternative is non-negative.

We define the following decision variables:

xjkl Binary variable, xjkl= 1 when compensation level l for meeting of

type j at location k is selected; xjkl= 0 otherwise.

Iv Binary variable, Iv= 1 if the company invests in a video-conferencing system;

(28)

The objective function becomes: min X,Iv M X m=1 wm K X k=1 Nk J X j=1 " OCjkm I X i=1 µik Ivfijk(X) + (1 − Iv) ˆfijk(X) # + Iv M X m=1 (wmF Cm),(3.1)

where the term

I X i=1 µik Ivfijk(X) + (1 − Iv) ˆfijk(X)

represents how meetings are distributed in the sample, given an investment decision. This sample distribution is a result from a questionnaire and is used to estimate the distribution of meetings of the whole population. Respondents are weighted for the number of meetings they participate in, i.e. µik= PINik

i=1Nik, ∀i ∈ {1, . . . , I}, ∀k ∈ {1, . . . , K}. To determine the total operational costs,

we multiply the distribution of meetings with the operational costs, sum over alternatives and multiply by the total number of meetings per year per location. Then the sum over locations is taken and finally the sum over al cost components is taken, while taking into account the weights for the cost components. The total fixed costs are determined by summing the fixed costs over the costs components and taking into account the weights for the cost components.

The objective function is subjected to the following constraints:

L X l=1 xjkl= 1 ∀j ∈ {1, . . . , J }, ∀k ∈ {1, . . . , K}, xjkl ∈ {0, 1} ∀j ∈ {1, . . . , J }, ∀k ∈ {1, . . . , K}, ∀l ∈ {1, . . . , L} Iv ∈ {0, 1}.

The first constraint specifies that for each alternative at a specific location one compensation level must be selected. The other two constraints specify the binary domain of the decision variables.

3.2.2 Specific model

For the specific problem the following parameters are added to the model:

Pl Price for compensation level l.

N CCjk Costs of non-chargeable hours for meeting of type j at location k.

T Cjk Travel costs for meeting of type j at location k.

CT Ejk Cost of technical equipment for meeting of type j at location k.

ECjk Environmental costs for meeting of type j at location k.

EVl Value employees attach to getting compensation level l.

¯

x∗_k Mean value employees attach to the availability of video-conferencing when having a

meeting at location k.

ICs Annual costs for investment in a single screen video-conferencing system.

ICd Annual costs for investment in a dual screen video-conferencing system.

cap Maximum number of meetings possible on a video-conferencing system during a year.

n Number of video-conferencing systems.

ns Number of single screen video-conferencing systems.

nd Number of dual screen video-conferencing systems.

a Fraction of single screen systems.

(29)

Mathematical model

j ∈ {1, . . . , 4} is specified by

j = 1: physical meeting, traveling by car, j = 2: physical meeting, traveling by public transport,

j = 3: tele-conferencing, j = 4: video-conferencing.

k ∈ {1, . . . , 3} is specified by

k = 1: Den Haag, k = 2: Utrecht, k = 3: Groningen.

m ∈ {1, . . . , 3} is specified by

m = 1: economic costs, m = 2: environmental costs,

m = 3: social costs.

The operational costs OCjkm are specified by

OCjk1 = L X l=1 (Plxjkl) + N CCjk+ CT Ejk+ T Cjk, ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, . . . , 3}, OCjk2 = ECjk, ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, . . . , 3}, OCjk3 = −Ivx¯∗k, ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, . . . , 3}

The economic operational costs for meetings of type j at location k consists of compensation costs, costs of non-chargeable hours, cost of technical equipment and travel costs. The environ-mental operational costs are simply given by the environenviron-mental costs of having a meetings of type j at location k. The social operational costs are the negative of the value employees attach to the availability of video-conferencing. Furthermore, this value is only included when investing in

video-conferencing systems. ¯x∗_k is specified in Section 6.4.1. At this point in the modeling process

it is unknown how employees value compensations. Therefore, the term w3g (X, EVl) is added

to the model, which includes an unknown function g that depends on the selected compensation

levels and the valuation of compensations. w3g (X, EVl) is specified in Section 6.4.2.

The fixed costs F Cmare given by

F C1= nsICs+ ndICd, F C2= 0, F C3= 0.

The economic fixed costs consists of the investment costs of video-conferencing systems. The firm can invest in single screen and dual screen systems. Fixed environmental costs are zero, since environmental costs of producing video-conferencing systems are included in the operational en-vironmental costs. The fixed social costs are also zero.

The number of video-conferencing systems n is specified by

n = dˆne, where ˆn = _cap1

3 X k=1 Nk I X i=1

µikfi4k(X). That is, demand for video-conferencing is satisfied.

I

X

i=1

µikfi4k(X) represents the fraction of video-conferences at location k. Taking the sum over all

meetings and dividing by the capacity of a system, gives the required number of systems. Video-conferencing systems can be single screen and dual screen systems. The share of single screen

systems is specified by ns= dane, where a ∈ [0, 1]. Therefore, the number of dual screen systems

nd is specified by nd= n − ns.

The optimization model becomes

(30)

subject to L X l=1 xjkl = 1 ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, . . . , 3}, n = dˆne, ns+ nd = n, ns = dane, a ∈ [0, 1] xjkl, ∈ {0, 1} ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, 2, 3}, ∀l ∈ {1, . . . , 4}, Iv ∈ {0, 1}, and where µik = PINik

i=1Nik ∀i ∈ {1, . . . , I}, ∀k ∈ {1, . . . , 3},

ˆ n = _cap1 3 X k=1 I X i=1 µikfi4k(X), OCjk1 = 4 X l=1 (Plxjkl) + N CCjk+ CT Ejk+ T Cjk ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, . . . , 3}, OCjk2 = ECjk ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, . . . , 3}, OCjk3 = −Ivx¯∗k ∀j ∈ {1, . . . , 4}, ∀k ∈ {1, . . . , 3}, F C1 = nsICs+ ndICd, F C2 = 0, F C3 = 0.

The following chapters describe the specification of fijk(X), ˆfijk(X), ¯x∗k, g (X, EVl) and describe

(31)

Chapter 4

Conjoint Analysis

This chapter describes how conjoint analysis is used to model choice behavior. First conjoint analysis is introduced and a motivation is given for the use of conjoint analysis. Subsequently we discuss how additive random utility models form the foundation of conjoint analysis. For implementing the conjoint analysis, the experimental design is determined. Then the estimation procedure and results of the conjoint analysis are presented.

4.1 Introduction and motivation

This section describes the basic concepts of (choice based) conjoint analysis and gives a motivation for its usage. Conjoint analysis supports valuation as well as describing behavior. Furthermore, conjoint analysis is more efficient than contingent valuation, and therefore conjoint analysis is used in this research.

4.1.1 Introduction to general conjoint analysis

Conjoint analysis is a measurement technique developed in the field of mathematical psychology and psychometrics in the early 1960s by Luce and Tukey (1964). This technique became a popular method for measuring buyers’ value trade-offs of products with multiple attributes and limited levels per attribute. The products are called profiles. Respondents are asked to evaluate various profiles by rating or ranking them or by choosing the most preferred profiles from given choice sets. From these evaluations, consumers’ relative importance of various attributes of products can be estimated and the desirability of new products can be determined. For a more detailed introduction to general conjoint analysis, see Green and Wind (1975).

4.1.2 Introduction to choice based conjoint analysis

In this research, employees have to choose between various types of meetings. Choice based

(32)

A combination of levels specifies the profile of a car. One profile might be gasoline, two seats,

red, Mercedes,e40,000, while an other profile might be diesel, four seats, red, Ford, e20,000. In

conjoint choice experiments it is assumed that each level of the attributes has a utility level. The total utility of a profile is found after summation of the utilities of the attribute levels. So offering a product with lower utility for a specific attribute level can be compensated by a high utility for

an other attribute level. For example when Mercedes is preferred to Ford, a Mercedes ofe40,000

may have the same level of utility as a Ford of e20,000. Choice based conjoint models can be

formalized by additive random utility models with random parameters, which are described in Section 4.2.

4.1.3 Motivation of using choice based conjoint analysis

Before we describe choice based conjoint analysis in more detail, a motivation for the use of the method is given. Note that we are interested in the distribution of meetings over the various types of meetings, the influences of stimuli on this distribution and the added value of the availability of video-conferencing systems as perceived by employees.

Added value for employees when investing in video-conferencing systems

Investing in video-conferencing systems creates social value when employees decide to make use of the systems. This added value can be determined by a willingness-to-accept (WTA) type of argument. By giving employees a compensation for not making use of video-conferencing, the demand for video-conferencing is reduced. The compensation required to let employees decide not to make use of video-conferencing, is an approximation of the added value of the availability of video-conferencing systems. In other words, for making video-conferencing unavailable, employees are willing-to-accept a particular level of compensation. The WTA value can be determined by conjoint analysis and contingent valuation. The methods are discussed below.

Contingent Valuation

Contingent valuation method (CVM) is a stated preference method to determine the value of var-ious kinds of goods and services. CVM can be applied by asking consumers open-ended questions to reveal their willingness-to-pay (WTP) or willingness-to-accept (WTA) value for a good or ser-vice. An other approach uses a closed-ended question that present a bid for a good or serser-vice. The consumer is asked whether or not the bid overstates the consumer’s WTA or WTP value for that product. Respondents can be confronted with several bids when using closed-ended questions. The WTA or WTP value of open-ended questions can simply be determined by calculating the mean of all respondents. For closed-ended questions this value can be determined by parametric and non-parametric methods. For a discussion of these methods, see McFadden (1994). Closed-ended questions are mainly used in contingent valuation studies, since open-ended questions often result in non-responses. McFadden (1994) states that open-ended CV studies give significantly different results than closed-ended CV studies. Given that there exists a true WTP value, it is unclear whether and which method gives the right value.

Conjoint Analysis

(33)

Conjoint Analysis

price on choices can be determined. From this price effect, the WTA or WTP value of a good can be determined. The methodology is discussed in Section 6.4.1.

Comparison of Contingent Valuation and Conjoint Analysis

Conjoint analysis has some potential advantages over contingent valuation, see Adamowicz, Box-all, Williams, and Louviere (1998). CA is based on the valuation of attributes and can be used to value situational changes, by evaluating behavior of consumers. Conjoint analysis can be used to predict behavior and for valuation of goods in various situations. This makes CA useful for our particular problem, where the distribution of meetings over various types of meetings is esti-mated for several situations and the added value of the availability of video-conferencing systems is investigated. Furthermore, in conjoint studies the value of goods can be expressed in monetary terms as well as in terms of other goods. In CVM studies more tasks are needed than in CA studies to retrieve the same amount of information from respondents. Stevens, Belkner, Dennis, Kittredge, and Willis (2000) reports that in CA respondents are encouraged more to explore their preferences and value trade-offs, since substitutes are made more explicit in CA studies. This may lead to a more accurate valuation. Besides, in CA studies respondents can express ambivalence or indifference directly, which may reduce non-responses or protest behavior.

From the perspective of neo-classical economic theory, CVM and CA should yield similar results. However, Stevens et al. (2000) reports that WTP values of most conjoint studies are larger than those of contingent valuation studies. For the study described in Stevens et al. (2000) no evidence is found for differences between results of CVM and CA studies. Adamowicz et al. (1998) finds that the results of CA are somewhat lower than the results of CVM. More research has to be done for testing the difference between outcomes of CA and CVM studies. For the moment we assume that both methods yield similar results. Since CA supports valuation as well as it supports describing behavior, and CA is more efficient than CVM, CA is used in this research.

4.2 Random utility models with random parameters

In this section the statistical model upon which conjoint analysis is based is described. A logit model that describes choice behaviour is derived. Choice based conjoint analysis is based on random utility theory. In additive random utility models respondents choose an alternative, from a choice set, with the highest utility. The utility of individual i for profile j in choice set k is specified by

Uijk = Vijk+ ijk, i ∈ {1, . . . , I}, j ∈ {1, . . . , J }, k ∈ {1, . . . , K},

where Vijk denotes the unknown utility component and ijk the random error. The unknown

utility component can be specified by Vijk = Xijkβi, where Xijk is a (1 × A) vector of variables

representing the characteristics of the jth alternative in choice set k for individual i, βiis a (A × 1)

vector of unknown utilities and A is the number of attributes describing the product. Since it is

likely that individuals have their own preferences, we let βi vary over individuals. Heterogeneity

can be assured by assuming that utilities follow a multivariate normal distribution, that is

βi∼ N (β, Σβ).

The vector βican be interpreted as population mean vector of utilities β and the individual specific

deviation from that mean ηi. The random utility model can be rewritten as

(34)

where ηi ∼ N (0, Σβ]). The introduction of random parameters has the property that the error

term Xijkηi+ ijk is correlated across alternatives and choice sets. That is

Cov[Xijkηi+ ijk, Xilkηi+ ilk] = Xijk0 ΣβXilk, j 6= l

and

Cov[Xijkηi+ ijk, Xijlηi+ ijl] = Xijk0 ΣβXijl, k 6= l.

In general utilities can not be observed. One can observe the actual choices made by a respondent.

The actual choices Yijk for each individual i in choice set k can be specified by:

Yik= j iff Uijk≥ Uilk, ∀l 6= j, l ∈ {1, . . . , J }. (4.1)

The probability of respondent i choosing profile j in choice set k is defined by

P (Yik= j) = P (Uijk≥ Uilk),

= P (Uilk− Uijk≤ 0),

= P (ilk− ijk≤ Vijk− Vilk, ∀l 6= j, l ∈ {1, . . . , J }).

When the errors ijk are independent of βi and iid type 1 extreme value, with density

f (x) = e−xexp[−e−x]

and distribution function

F (x) = exp[−e−x],

it can be shown that conditional on βi

P (Yik= j|βi) = eXijkβi X l∈{1,...,J } eXilkβi , (4.2)

which is the standard multinomial logit model. Since βi is random and can not be observed, the

conditional probability can not be found. To find the unconditional probability, the integral of

P (Yik= j|βi) over all possible values of βi must be taken. This yields

P (Yik= j) = Z _eXilkβi X l∈{1,...,J } eXilkβi φ(βi|β, Σβ)dβi, (4.3)

where we have a multidimensional integral and φ(βi|β, Σβ) denotes the multivariate normal

den-sity for βi with mean β and variance/covariance matrix Σβ. The integral has no closed form

(35)

Conjoint Analysis

4.2.1 Independence of irrelevant alternatives

Because the error terms are correlated across alternatives, the random parameters logit model does not suffer from independence of irrelevant alternatives (IIA) assumption. The standard logit model with non-random parameters does assume independence of irrelevant alternatives. The IIA assumption implies that the relative probability of two profiles does not depend on the attributes describing a third profile. For example, when profile A, B, C and D all have probabilities of 0.25 to be chosen, removing D from the choice set results in equal probabilities of 0.33 for A, B and C. Adding profile E to choice set {A,B,C} and capturing a probability of 0.40 to be chosen results in probabilities of 0.20 for A, B and C. So changing one profile, while keeping others constant, results in the same relative probabilities of the unchanged profiles. This assumption is strong and it is likely that IIA does not hold when two profiles are close substitutes. Take for example the red bus/blue bus example from the literature. Consider a choice problem where individuals choose between two modes of transportation: car and red bus. Assume that both transportation modes are chosen with equal probability. Adding a red bus as third alternative to the choice set and assuming that individuals do not care about the color of the bus, we expect a choice probability of 0.5 for the car, and a probability of 0.25 for both of the buses. However, the IIA assumption implies that the three modes of transportation are chosen with equal probability of 0.33. The IIA assumption fails to take into account perfect substitutes. In this research we realize that traveling by car or public transport to a meeting may be close substitutes and similarly, video-conferencing and tele-conferencing may be close substitutes. However, by using the random parameter logit model, we do not have have to assume IIA for the aggregated model. Only at the individual level IIA is assumed.

Is the IIA assumption at the individual level a problem? It is a problem when probabilities are predicted and close substitutes are added or removed to or from a set of alternatives. In the research we are interested in the allocation of meetings when video-conferencing is available and when video-conferencing is not available. If video conferencing is a close substitute for tele-conferencing or having a face-to-face meeting, then we could have a problem. The predicted market shares would not reflect the actual behaviour of employees. To check whether video-conferencing has a close substitute, a hold-out task is included in the questionnaire. A hold-out task is a fixed task that every respondent has to complete. This hold-out task is not used to estimate the model, but it is used to validate the estimated model. When the estimated model gives similar results as the results of the hold-out task the model is accepted. To test the IIA assumption we include a hold-out task without video-conferencing available. When the results of the hold-out task differ significantly from the results of the model, the hypothesis that the IIA assumption holds, is rejected. If IIA does not hold, we can not predict how employees react on stimuli when video-conferencing is not available.

4.2.2 Estimation of random parameter logit model

The random parameters logit model can be estimated by Bayesian estimation methods (see Cameron and Trivedi, 2005, chapter 15.7.2). We developed a hierarchical model, where at the higher level individuals’ utilities (also called part worths) are described by a multivariate normal distribution. At the lower level, a multinomial logit model is defined to describe an individual’s probability to choose a particular alternative, given his/her part worths. The parameters to be

estimated are the individual vectors of part worths βi, the vector β and matrix Σβ representing

the mean and covariance matrix of the distribution of part worths, where each βi is considered

to be a parameter along with β and Σβ. To estimate the parameters, the Bayesian method uses

(36)

4.3 Experimental design

We have described a choice model and an estimation procedure to predict choices given a data set containing choices from choice sets. We continue by describing the data collection method. As stated earlier, the choice data is gathered by a questionnaire. To set up a questionnaire, several steps have to be taken.

First important attributes are selected. The choice of having a meeting of a particular type is modeled. This choice can be influenced by several factors, e.g. location, number of participants, elapsed time since previous meeting, nature, expected length, workload, perception on the quality of a certain type of meeting, weather conditions, etcetera. A meeting at a distant location results in long traveling time when choosing for a physical meeting. When the length of the meeting is several hours, the traveling time may not be a burden. However, when an employee has a busy agenda, traveling to a meeting may be a burden. When participants have not seen each other for a long time, they might prefer a physical meeting. A face-to-face meeting might also be preferred (or not be preferred) when a large number of people are participating in a meeting. The nature of a meeting can also influence the choices made. For example, for a job interview and a kick-off meeting a different type of meeting may be appropriate than for a regular meeting. Furthermore, employees may have perceptions of the quality of meetings of a particular type. These perceptions can be either right or wrong. In any case, perceptions do influence the actual choices made. Also weather conditions can influence the choices made. An employee deciding on a type of meeting can take all these factors in consideration, and even more.

In this research we do not focus on all the factors that influence the choices. Trade-offs made by employees are considered as a black-box. The questionnaire is only controlled for location of meetings, since this factor has great influence on the costs for a company. By considering a project with ten meetings at a particular location, a respondent can give its own interpretation of the nature, length, number of participants, etcetera for the meetings. By asking to allocate ten meetings over the four different types of meetings, a respondent can make its choices based on his own interpretation of the project. Besides types of meetings, the product profile consists of stimuli. We are interested in the effects of stimuli on the allocation of meetings, since we are trying to find an allocation of meetings which maximizes objectives, and because the added social value of video conferencing is investigated. Stimuli are given in the form of compensations. The two attributes in the design are (1) type of meeting and (2) compensation.

Analyzing the Eﬀects of Monetizing Intangibles in Decision Analysis A Case Study for Video-Conferencing

Analyzing the Effects of Monetizing Intangibles in Decision Analysis

A Case Study for Video-Conferencing

Analyzing the Effects of Monetizing Intangibles in Decision Analysis

A Case Study for Video-Conferencing

Management summary

Preface

Contents

Chapter 1

Introduction

1.1

Incorporating qualitative aspects in decision making

1.2

Outline of thesis

1.3

Case study

1.4

TNO

Chapter 2

Problem formulation

2.1

Context of research

2.1.1

Corporate social responsibility

2.1.2

Decision analysis

2.2

Research objective and research questions

2.2.1

Introduction to the case study

2.2.2

Meetings and perceptions on meetings

2.2.3

Description of objectives in the case study

2.3

Methodology

2.3.1

Formulating a mathematical model

2.3.2

Valuation of objectives

2.3.3

Value trade-offs between objectives

2.3.4

Solving the decision problem

2.4

Level and scope of research

2.5

Literature review on quality of meetings

Chapter 3

Mathematical model

3.1

Model description

3.1.1

Input and output

3.1.2

Objective and constraints

3.1.3

Graphical representation of the relations in the model

3.2

Model specification

3.2.1

General model

3.2.2

Specific model

Chapter 4

Conjoint Analysis

4.1

Introduction and motivation

4.1.1

Introduction to general conjoint analysis

4.1.2

Introduction to choice based conjoint analysis

4.1.3

Motivation of using choice based conjoint analysis

4.2

Random utility models with random parameters

4.2.1

Independence of irrelevant alternatives

4.2.2

Estimation of random parameter logit model