SELECTING PROJECTS FOR VENTURE CAPITAL FUNDING:

A MULTIPLE CRITERIA DECISION APPROACH

Gina Beim, P.E.MCDA Consulting LLCgina.beim@mcdaconsulting.com

Moren Lévesque, PhDSchulich School of BusinessYork University

Venture Capitalists & their Decisions

Selecting businesses for investment 3 broad criteria:

– quality of management – unique product or market opportunity– potential for capital appreciation

Evaluation process: – objective information gathering and analysis– intuition, gut feeling and creative thinking

Modeling the VC Decision Process

Direct criteria weighting with questionnaires (MacMillian et al., 1985 and Fried et al., 1993)

Conjoint analysis (Muzyka et al., 1996, Zacharakis and Meyer 1998, Shepherd, 1999, Riquelme and Rickards, 1992) -

Actuarial models (Zacharakis and Meyer, 2000)

UTA (Utilité Additive) models (Siskos and Zopounidis, 1987)

Modeling the VC Decision Process

Conjoint Analysis: acknowledges the multiplicity of criteria; relative weights inferred; limited in criterion rating; utilizes hypothetical evaluation as initial point

Actuarial bootstrapping models and UTA: related to Multi Attribute Value Theory (MAVT); utilize decision maker’s real past evaluations as initial point.

Shepherd and Zacharakis, 2002: A call for more than reproducing the investment selection process, and instead for the use of decision aids in the venture capital world.

MCDA in Financial Decision Making

Investment portfolio selection (Bouri, Martel and Chabchoub, 2002),

Extension of credit (Matsatsinis, 2002) Foreign direct investment (Doumpos,

Zanakis and Zopounidis, 2001) Several papers presented in this conference

Promising New Field of Application for MCDA: VC Portfolio Selection

Bridges gap between official and de facto policies: helps VCs understand and express what policies are; incorporates policies into decision model.

Interactive sensitivity analysis: brings aspects not previously considered to forefront.

Belton and Stewart (2002: 283): “most memorable interventions in organizations have been those in which the multicriteria analysis has brought about a strong challenge to the decision making group’s intuition”.

The JumpStart Fund

Created by business and academic leaders to provide start-up capital to companies headquartered in Northeast Ohio.

$2.3 million fund Based at Case Western Reserve University

between 2001 and 2003. In 2004 became part of a larger organization.

Until 2003, a typical JumpStart investment amount was in the range of $200,000.

9 Business Plans in our Case Study

Dental device E-commerce facilitation Human resources tool Management software Market research tool Media company Medical device Pharmaceutical Supply chain management software

Modeling and Analysis

Multi Attribute Value Theory – Logical Decisions® software.

Criteria developed in interactions with JumpStart fund manager.

Combination of top-down and bottom-up structuring techniques.

Fund manager encouraged to avoid criteria redundancy, lack of independence, and extreme complexity while being comprehensive and sensitive to criteria relevance.

Model Structure

Overall goal: “Selecting the Best Businesses to Fund”.

4 sub-goals: “Management and Governance”, “Feasibility of Proposition”, “Market Considerations” and “Return on Investments”.

10 lower level (measurable) criteria. Criteria critically evaluated against entrepreneurship

literature and practice.

Founder's track record

Measure

Quality of Board

Measure

Quality of Management

Measure

Management and Governance

Goal

Realistic Approach to Financing

Measure

Well thought out milestones

Measure

Feasibility of Proposition

Goal

First Mover?

Measure

Potential Market Size (billion US$)

Measure

Proprietary Techonology / Patent Protection

Measure

Market Considerations

Goal

Exit Opportunities

Measure

Time to Achieve Profitability

Measure

Return on Investment

Goal

Successful venture

Goal

Hierarchy of Criteria

for Business

Plan Evaluation

Business Plan Ratings

Ratings based on information contained in the business plans.

Performance assessed on an interval scale of measurement containing minimum and maximum local reference points.

Group of business plans being analyzed was representative of the universe of plans targeted by JumpStart: global and local reference points coincided.

Fund manager had choice of categorical or ordinal scales. Mostly chose a subjective categorical scale.

Business Plans Ratings

Business PlanExit Opportunities

First Mover?

Founder's track record

Potential Market Size (billion US$)

Proprietary Techonology / Patent Protection

Quality of Board

Quality of Management

Realistic Approach to Financing

Time to Achieve Profitability (years)

Well thought out milestones

dental deviceAcquisition likely yes High 1 Patent protected

No board mentioned High

Highly realistic 3

Well thought out

e-commerce facilitationNo exit opportunity yes Medium 0.5 patent pending Medium Medium

Financing not mentioned 3.45

No Milestones mentioned

human resources toolAcquisition likely no Medium 3.3 No protection High High

Highly realistic 1

Well thought out

management softwareAcquisition likely no High 3.6 No mention

No board mentioned Medium

Somewhat realistic 0

Well thought out

market research toolAcquisition likely no Medium 5.9 No protection High Medium

Highly realistic 1.21

Well thought out

media companyNo exit opportunity yes Low 0.1 No mention

No board mentioned Low

Highly realistic 1

Somewhat realistic

medical deviceAcquisition likely yes Low 4.8 patent pending High Medium

Highly realistic 5

Well thought out

pharmaceuticalAcquisition likely yes Medium 3.375 patent pending

No board mentioned Medium

Financing not mentioned 1

Somewhat realistic

supply chain management software

Acquisition likely no Low 15 No mention High Medium

Somewhat realistic 1

Well thought out

Probabilistic Assessment

Point estimates of discrete probabilities of each event or expected values of uniform distributions between the upper and lower estimates as mentioned in the business plans.

Probabilistic ratings incorporated in the analysis.

Subjective probability estimates. Elicitation avoided cognitive biases.

Weight Elicitation

Swing-weight for the lower level criteria. For higher level goals, the fund manager felt

very strongly that all goals should have equal weights. We revisit this proposition in the sensitivity analysis.

Value Function Elicitation

Direct assessment for criteria with only a few possible discrete values.

Value functions for the two criteria modeled by continuous variables were assessed with the aid of software graphical tools.

Additive value function to aggregate the value functions for each criterion: very intuitive, widely used in practice, and mathematically sound.

Value Function for “Time to Achieve Profitability”

Utility

Time to Achieve Profitability (years)

1

0

0. 5.

Selected Point -- Level: Utility:3.11111 0.886154

Alternativesupply chain management software

dental device

human resources tool

medical device

market research tool

management software

pharmaceutical

media company

e-commerce facilitation

Value 0.824

0.777

0.766

0.729

0.660

0.637

0.542

0.349

0.261

Ranking for “Successful Venture” Goal

Results and Sensitivity Analysis

Sensitivity to outcome of probabilistic assessment. Sensitivity to weights. Ranking of top 5 alternatives very robust; rank

reversal only between “medical device” and “market research tool”.

Equal weights for the 4 higher level goals revisited. Top ranked alternatives insensitive to weight variation in those goals.

Discussion

JumpStart fund manager selection corresponded to the 4 highest ranked businesses. These had exhibited considerable robustness to variations in weights or probabilistic ratings.

Confidence of venture capitalists in the methodology– Increased for JumpStart manager, but did not prompt

reconsidering the fund decision process.– Consultations with other VCs revealed cautious interest. – Zacharakis and Meyer’s (2000): VCs reluctant to use decision

aids.

Potential Contributions

Addresses Zacharakis and Meyer (2000) suggestion that models better reflect the “needs and beliefs” of each individual firm.

Improves dichotomous attributes from conjoint analysis of Shepherd et al (2000).

Gives VCs feedback on decision processes called for by Shepherd and Zacharakis (2002).

Allows for greater flexibility than other models in scales choice. Captures a VC’s uncertainty. Minimizes cognitive biases of seasoned VCs. Encourages inexperienced VCs to engage in systematic rating

and critically examine results via sensitivity analysis.

Limitations Zacharakis and Meyer (2000): improvement = selecting higher %

of successful business plans than the VCs. We cannot make that claim, but we can claim better educated, more transparent and more thought out decisions.

We cannot ascertain elimination of bias but we minimize them by structuring the interview encouraging fund manager to think carefully about each probabilistic estimate and conducting sensitivity analysis

Fund manager preferences may not be entirely consistent and rational, but sensitivity analysis accounts for this and allows for a reevaluation of preferences.

VCs who report taking an average of only 8 to 12 minutes to evaluate a business plan may resist MCDA, but our fund manager did not share that evaluations could be so quick (12 minute is an average).

Conclusions

MCDA: goal is not to replace or outperform VCs, but to improve their decisions by shedding light into the complexities of the choices they face and minimizing their cognitive biases. Better results may be a natural consequence.

Future research: – Methodologies and processes that facilitate MCDA acceptance

by VC community. – How to conduct interviews in a manner that at the same time

minimizes errors in judgment, maximizes the comfort level of the VC, and retains all the necessary validity conditions for the construction of a mathematically rigorous MCDA model.

Thank you.

Questions or Comments?