Bernd Skiera, Martin Spann
Opportunities of Virtual Stock Markets to Support New Product Development
1 2 3 4
Introduction........................................................................................................................ 3 Opportunities of the Internet for New Product Development..................................... 4 Virtual Stock Markets and Their Use in New Product Development Stages............. 5 Empirical Study.................................................................................................................. 8 4.1 Design of the Study .................................................................................................. 9 4.2 Forecast Accuracy..................................................................................................... 9 4.3 Performance Compared to Expert Judgments .................................................... 11 4.4 Factors Influencing Forecast Error ....................................................................... 12 Summary and Conclusions............................................................................................. 13 References ......................................................................................................................... 14
5 6
Bernd Skiera, Martin Spann
2
Opportunities of Virtual Stock Markets to Support New Product Development
1
Introduction
Although the Internet has not fulfilled all expectations of the stock markets, it has been adopted by many consumers and companies all around the world. Recent studies report a 53.5% adoption rate among consumers in Germany (Eimeren et al. 2003) and a 60% adoption rate among consumers in the United States (Rayport and Jaworski 2001). In Western Europe alone, more than 250 Mio. consumers are expected to have access to the Internet in 2005 (European Information Technology Observatory 2004). Worldwide business‐to‐business electronic commerce is expected to be greater than 1 Trillion USD in 2003 (Rayport and Jaworski 2001). This wide acceptance of the Internet alters prod‐ uct development (Dahan and Hauser 2002). Yet, new product development still re‐ mains difficult and costly (Di Benedetto 1999, Brockhoff 1999). The flop rates of newly launched products remain high over the years, often surpassing 50% (Urban and Hauser 1993). Hence, even small improvements in the new product development process can have a major effect on companiesʹ profits and competitive advantage if this flop rate is reduced. Therefore, new methods to improve new product development are of high relevance for companies. Virtual stock markets could be such a method. They have recently gained much attendance (Spann and Skiera 2003c, Wolfers and Zitzewitz 2004, Polk et al. 2003) and even the Pentagon considered virtual stock markets as a tool to better forecast economic and political stability (Hulse 2003). The basic idea of virtual stock markets is to make future events or market situations expressible and tradable through virtual stocks (Spann and Skiera 2003b, Forsythe et al. 1992). Thereby, the cash divi‐ dend (payoff) of such shares of virtual stocks depends on a particular outcome, e.g., the success of a software development project (Ortner 2000), product sales (Plott 2000), goodness of new product concepts (Chan et al. 2002) or new product sales (Spann and Skiera 2003a). Studies show that virtual stock markets might have the potential to support new product development successfully (Spann and Skiera 2003a). Yet, none of those studies provided a comprehensive analysis about the particular stages of new product development that might be supported. In addition, little is known about the factors that influence the forecasting accuracy of virtual stock markets. Therefore, the aim of this paper is to analyze the opportunities of virtual stock markets to support new product development and to empirically determine factors that influ‐ ence the forecasting error of virtual stock markets. For that reason, we analyze in Sec‐ tion 2 the impact of the Internet on new product development. Section 3 describes virtual stock markets and their opportunities to support the different stages of the new product development process. In Section 4, we describe an empirical study that uses virtual stock markets to forecast the success of new products, compare forecasting accuracy with those of expert judgments and analyze the factors that influence forecast accuracy. Section 5 summarizes the implications of the paper.
3
Bernd Skiera, Martin Spann
2
Opportunities of the Internet for New Product Development
Brockhoff (1999) proposes to distinguish five stages of the new product development process, namely (i) idea generation and screening, (ii) development of product con‐ cepts, (iii) research & development, design and engineering of product prototypes, (iv) product testing and (v) product launch. Others share this point of view (Urban and Hauser 1993, Dahan and Hauser 2002). Therefore, we use these five stages to structure the different ideas that have been proposed to support new product development (see Figure 1).
Figure 1:
Opportunities of the Internet to support stages of new product development
Stages of New Product Development
Idea generation & screening
Opportunities of the Internet to Support NPD
• Analysis of Online Communities and Newsgroups • Web-based Creativity Contests • Web-based Lead User Identification • Web-based Conjoint Analysis
Product Concepts
Design & Engineer Product Testing Product Launch
• Web-based Design Collaboration Tools to link multinational development teams • Web-based Concept Testing (e.g., communities)
• Promotion via Communities • Product Websites (e.g., movies)
Figure 1 provides an overview of the opportunities of the Internet to support new product development. The idea generation and screening stage can be supported in several ways by the Internet. Online communities and newsgroups can be systemati‐ cally analyzed for new product ideas. In addition, creativity and idea generation con‐ tests can be easily organized via the Internet (Ernst et al. 2004). Thereby or in connec‐
4
Opportunities of Virtual Stock Markets to Support New Product Development
tion with an online survey, a company can try to identify lead users, which it can then use as a source for new product ideas (Urban and Von Hippel 1988, Brockhoff 2000). In the product concept stage, consumer preferences for different new product concepts can be evaluated via web‐based preference elicitation tools such as conjoint analysis. Thereby, the presentation of new product concepts as well as the preference elicitation method can be conducted completely online, saving time and money, as well as mak‐ ing use of the graphic and audio capabilities of the world wide web to depict virtual products and product features ((Dahan and Hauser 2002, Dahan and Srinivasan 2000, Ernst and Sattler 2000). In addition, the computational capabilities of the Internet al‐ low to dynamically adapt web‐pages in real time (Toubia et al. 2003). Web‐based design collaboration tools, such as computer aided design (CAD) and computer aided manufacturing (CAM), linked to a companyʹs knowledge manage‐ ment system, can support interaction between multi‐regional and multi‐national R&D‐ teams. Further, such tools and online communities in a company intranet as part of its knowledge management system can enhance collaboration between different depart‐ ments engaged in the design and engineering stage of the new product development process (Grover and Davenport 2001). Product prototypes can be tested among an online community as part of the product testing stage (Panten et al. 2001). Web‐based preference elicitation tools can be applied at this stage as well. The launch of a product can be supported by specific product websites (e.g. for mov‐ ies), which inform consumers about the product and thereby help to reduce buyer uncertainty for fairly new products. Further, new products can be promoted via online communities and newsgroups (Albers et al. 1998). In addition, product placements in Online games can provide a new opportunity to promote products.
3
Virtual Stock Markets and Their Use in New Product Development Stages
The idea of virtual stock markets is to bring a group of participants together via the Internet and let them trade shares of virtual stocks. These stocks represent a bet on the outcome of particular future events and their value depends on the realization of these events. Once the occurrence of the particular event is known, each share of virtual stock receives a cash dividend (payoff) according to that particular event (e.g., $1 for each unit sold). As those ʺstocksʺ are actually securities because their terminal values are contingent upon the outcome of an uncertain event, some authors use the label ʺsecurityʺ (Dahan and Hauser 2002, Chan et al. 2002). However, in accordance with the
5
Bernd Skiera, Martin Spann
major part of the literature dealing with virtual stock markets (e.g., Forsythe et al. 1992, Forsythe et al. 1999, Spann and Skiera 2003b), we use the denomination ʺstocksʺ because that makes the concept easier to understand for the major part of the partici‐ pants. Such types of virtual stock markets were first applied in the form of a political stock market to predict the outcome of the Bush vs. Dukakis US presidential election in 1988 (Forsythe et al. 1992). Afterwards, virtual stock markets were used to predict the re‐ sults of many other elections (Forsythe et al. 1999, Spann 2002). Later on, researchers started to apply virtual stock markets to solve business problems. Ortner (2000) uses virtual stock markets for the success of a software development project, Plott (2000) for predicting product sales and Spann and Skiera (2003a) for forecasting new product sales. Most recently, the Pentagon intended to use virtual stock markets to derive fore‐ casts concerning foreign policy events, e.g., a coup dʹétat in certain countries (Polk et al. 2003, Spann and Skiera 2003c). The basic idea behind virtual stock markets is that the price of one share of a virtual stock should correspond to the virtual stock marketʹs aggregate expectations of the event outcome because participants of the virtual stock markets use their individual assessment of the particular event to derive an individual expectation of the cash divi‐ dend of the related share of virtual stock. According to the Hayek hypothesis, the market mechanism should be the best way to aggregate the individual assessments, because the price mechanism on a competitive market is the most efficient instrument to aggregate the asymmetrically dispersed information of market participants (Hayek 1945, Smith 1982). Virtual stock markets can be used as an information gathering tool to support new product development. The different opportunities of virtual stock markets to provide market intelligence in the new product development process are displayed in Figure 2, again using the distinction into five stages proposed by Brockhoff (1999). In the idea generation and screening stage an online community can be created, which is organized around existing products that are traded on a virtual stock market. One example is the Hollywood Stock Exchange (www.hsx.com) that performs a virtual stock market on the success of new movies and contains a major virtual community dealing with movie related topics. Thus, trading on the virtual stock market stimulates consumers to express and discuss new product ideas as well as new product success factors in the online community. The systematic analysis of this community can pro‐ duce new product ideas (Ernst et al. 2004). Further, participants of this virtual stock market can be analyzed in order to detect lead users (Spann et al. 2003). In the product concept stage, a virtual stock market can try to assess consumersʹ ag‐ gregated preferences for different new product concepts taken up ideas that have been proposed by Chan et al. (2002). One major problem of the design applied by Chan et al. (2002) is that the payoff value of stocks and thus the assessment of product concepts
6
Opportunities of Virtual Stock Markets to Support New Product Development
could be biased by a self‐fulfilling prophecy (Spann and Skiera 2003a). The reason is that Chan et al. (2002) use the final price in the stock market as payoff value. Spann and Skiera (2003a) propose a design modification of a payoff based on the results of two parallel experimental groups that might solve this problem if each groupʹs final stock price is used as the payoff for the stock prices of the other group. At the design and engineering stage, different design and development solutions can be evaluated at a virtual stock market on a companyʹs intranet. Thereby, the assess‐ ments on the feasibility and efficiency of different construction and manufacturing solutions can be traded by one or several R&D‐teams. Further, the inclusion of mem‐ bers for the marketing department as traders can add market‐related information. Virtual stock markets might especially be beneficial in such situations because the aggregation of the individual estimates will not be biased due to different positions in a companyʹs hierarchy (Spann 2002). Spann and Skiera (2003a) show in a different context that even virtual stock markets with only 12 participants are large enough to get good results.
Figure 2: Opportunities of Virtual Stock Markets to Support New Product Development
Stages of New Product Development
Idea generation & screening
Opportunities of Virtual Stock Markets to Support NPD
• Analysis of Online Community centered around product related VSM • Lead User Identification via VSM
Product Concepts
Design & Engineer Product Testing Product Launch
⎫ ⎪ ⎪ • Concept & Product Testing via VSM ⎬ • Combination of VSM and traditional ⎪ marketing research methods ⎪ ⎭
• Pre-Launch Forecasting • Identification of Target Groups
Product prototypes can be tested in a virtual stock market so that participants can trade their assessments on the market success of these different prototypes (Chan et al.
7
Bernd Skiera, Martin Spann
2002). Thereby, additional information could be elicited by combining a virtual stock market with traditional survey and focus group methods on the same set of consum‐ ers, because trading in the virtual stock market can stimulate consumers to focus on the subject and quantify their assessment of market success (Spann and Skiera 2003b). Virtual stock markets can be used for pre‐launch forecasting of a productʹs market success. Such forecasts are very useful for a company in order to optimize their prod‐ uct‐launch related marketing instruments. For example, a movie studio can use this information to decide on promotions and advertising related to the movieʹs release. Movie exhibitors can plan on whether to display the movie in large or small theatres. (Spann and Skiera 2003b). Further, an analysis of tradersʹ portfolios and trading behav‐ ior might be useful for the analysis of target groups (Spann and Skiera 2003a). Compared to other knowledge gathering techniques applicable in the new product development process, virtual stock markets offer the following advantages (Spann and Skiera 2003b, Dahan and Hauser 2002): First, they allow for an almost real‐time reac‐ tion of stock prices to additional information and, hence, a very quick prediction of the impact of that information on future market situations. Second, it does not burden the researcher with the task of weighting and aggregating different expert judgments as this is achieved by the trading mechanism implemented in the virtual stock market. Participants, for example, weight their assessments by the volume and the price of the purchase or sale order they place or accept. Third, once established, a virtual stock market can operate at rather moderate operating costs, e.g. for repeated new product concept tests. Fourth, a virtual stock market provides participants with an incentive to reveal their true assessments (Forsythe et al. 1999), if an adequate remuneration is properly linked to the participantsʹ performance on the virtual stock market. Hence, whereas many consumer surveys remunerate consumers for their participation at a survey, a virtual stock market usually remunerates participants for their successful participation (Spann and Skiera 2003b, Dahan and Hauser 2002). Wertenbroch and Skiera (2002) show, for example, that consumersʹ willingness‐to‐pay differs signifi‐ cantly according to the incentive structure being provided. Finally, participants in a virtual stock market might have more fun than their counterparts partaking in con‐ sumer or expert surveys (Dahan and Hauser 2002).
4
Empirical Study
The goal of the following empirical study is to analyze the use of a virtual stock mar‐ ket to predict the success of new products prior to their launch. Thereby, we analyze the feasibility, the forecast accuracy and the factors influencing forecast accuracy of a virtual stock market to predict the success of new products, namely the success of
8
Opportunities of Virtual Stock Markets to Support New Product Development
movies in Germany. Movies face high financial stakes for production and marketing, a significant failure rate, and rather unstable market conditions. (e.g., Sawhney and Eliashberg 1996 or Eliashberg et al. 2000). Hence, we look at a virtual stock market that has been used as a pre‐launch forecasting tool in the product launch stage.
4.1
Design of the Study
We conducted the movie exchange (www.CMXX.com) seven times for the prediction of movies, using our own virtual stock market software (the first round also included the chart position of 11 pop music singles in Germany which we omit for our analysis). We conducted a virtual stock market for the prediction of the box‐office success (num‐ ber of visitors) of movies in Germany. During the seven rounds of CMXX, virtual stocks for ten to fifteen movies were traded in each round. In total, virtual stocks were traded for eighty‐one movies. At the end of CMXX, each share of movie stock received a cash dividend (payoff) according to the total number of visitors of the respective movie in Germany until the end of the specific round. Prices were limited to $3,000 (virtual) in the first round, and $3,500 (virtual) in the following rounds for movie stocks, considering that more than 3,000,000 and 3,500,000 movie visitors were unrealistic in Germany. In the first round, CMXX provided non‐ monetary incentives in the form of a ʺGolden Recordʺ and ten music CDs for the par‐ ticipant with the highest portfolio value, five and three music CDs respectively for the participants with the second and third highest portfolio values. Four sets of movie merchandise were given to randomly chosen participants ranking fourth to one hun‐ dredth according to final portfolio value. In the second to seventh rounds, the partici‐ pant with the highest portfolio value in each round received an annual ticket for a large German movie exhibitor; the participants with the second and third highest portfolio value received ten free movie tickets and a set of movie merchandise, respec‐ tively. Table 1 provides an overview of the design of the movie exchange.
4.2
Forecast Accuracy
The price of a share of a movie stock represents a prediction of the number of visitors for the selected movie up until the end of the specific round. Thus by multiplying the stock price with 1,000, the forecast of a movieʹs number of visitors can be easily de‐ rived.
9
Bernd Skiera, Martin Spann
Table 1:
Step
Design of the Movie Exchange
Decisions Forecasting the number of movie visitors in Germany Payoff function: Movie visitors in Germany: 1 virtual Euro per 1,000 visitors of a movie Duration: First round: 22 January – 5 February 2001; [Second to seventh rounds: Duration of one month each between May and October 2001] Open to the public; participants can join at any time
Choice of Forecasting Goal
Incentives for Participation and Information Revelation
Composition of Initial Portfolios / Endowment: Endowment of 100 shares of each type of movie stock and $500,000 [$250,000] (virtual) per participant Provision of loans up to $500,000 [$250,000] (virtual) at no interest rate per participant Remuneration / Incentive Mechanism: Nonmonetary rewards Rank-order tournament: Rewards for participants with the highest, second highest and third highest increase in (virtual) portfolio value (annual movie ticket, 10 free movie tickets, movie merchandize) Time interval: Whole virtual stock market duration Incentives not based on performance: First round: Lottery for four rewards among participants ranked fourth to one hundredth
Financial Market Design
Double auction trading mechanism with open order book Trading times: Twenty-four hours a day, seven days a week No short trading Order types: Limit and market without temporal restriction No position limits, maximum price limits of 3,000 [3,500] for movie stocks No trading fee
In each round, the movie exchange attracted around fifty actively trading participants. The forecasts derived from CMXX used the price of the last trade of a specific type of stock before trading was stopped at the end of a specific round. CMXX faced the prob‐ lem that it included movies with very few visitors and presumably little information available among the participants (e.g., the movie ʺalthan.comʺ had only 20,000 visitors compared to 2,296,000 visitors for ʺUnbreakableʺ). Consequently, forecast accuracy for
10
Opportunities of Virtual Stock Markets to Support New Product Development
the less publicized movies below 100,000 visitors was rather bad with an absolute percentage error of above 100% each (see Table 2).
Table 2:
APE Movie 1 Movie 2 Movie 3 Movie 4 Movie 5 Movie 6 Movie 7 Movie 8 Movie 9 Movie 10 Movie 11 Movie 12 Movie 13 Movie 14 Movie 15 MAPE Median Min Max
Forecast Error of Movie Exchange
Round 1 Round 2 Round 3 Round 4 Round 5 Round 6 Round 7 Overall .024 2.776 .332 .190 .057 .141 9.667 .705 3.516 .630 .250 26.889 .232 .031 .570 .307 .222 .170 2.784 .005 3.839 .263 13.318 .118 .133 5.507 1.663 .332 .024 9.667 3.626 .263 .005 26.889 5.589 .378 .061 49.000 .863 .621 .050 2.824 1.109 .170 .010 5.600 .733 .220 .012 5.000 .687 .190 .013 3.895 2.119 .263 .010 49.000 .096 .061 .400 1.667 .123 .208 .378 7.214 .074 2.258 49.000 .050 .198 .130 .956 2.824 .055 .407 .297 .835 2.390 1.041 1.174 .048 .010 .032 .170 .333 .102 5.600 .153 .701 3.115 1.935 .012 5.000 .252 .018 .053 .189 1.000 .250 n.a. .080 n.a. .479 .030 .028 .505 .031 .190 n.a. .281 2.049 .040 .500 .013 3.895
Bold print: Movies having over 100,000 visitors. n.a.: Movie release postponed. APE: Absolute Percentage Error.
4.3
Performance Compared to Expert Judgments
The performance of the CMXX results is compared to corporate expert predictions from the management of a large German movie exhibitor that we were able to collect for the first two rounds but not for additional rounds (see Table 3). We compare the predictions of CMXX directly. The expert predictions were provided approximately one week before the end of each round of CMXX and were not made available to the participants of CMXX. The CMXX hit rate in the first round was six out of ten for mov‐ ies (for the eleventh movie the CMXX prediction and the expert prediction were iden‐ tical). In the second round, the CMXX hit rate was eleven out of fifteen in comparison to the expert predictions from the movie exhibitor. Table 3 compares the Mean Abso‐
11
Bernd Skiera, Martin Spann
lute Percentage Error (MAPE) of CMXX to that of the expert predictions for movies having over 100,000 visitors. The forecasts of CMXX are significantly better than those of the experts, indicating that either CMXX performed well and/or that the experts performed poorly.
Table 3:
Instrument Round 1* Round 2*
Comparison between Predictions of CMXX and Experts (Movies > 100ʹ visitors)
CMXX: MAPE 13.83% 20.50% 18.59% Experts: MAPE 47.46% 115.73% 96.20% CMXX % improvement (pa) value) 70.86% (.331) 82.29% (.010) 80.68% (.005)
Round 1+2*
a)
Percentage of improvement of CMXX over alternative expert judgments: = [MAPE Expert – MAPE CMXX] / MAPE Expert (two-tailed paired t-test for difference)
* Movies having over 100,000 visitors. MAPE: Mean Absolute Percentage Error.
4.4
Factors Influencing Forecast Error
The results of our empirical study demonstrate, that virtual stock markets can some‐ times produce rather weak results. Therefore, it is important to derive factors which can indicate the expected forecast accuracy of a virtual stock market. In this section we will analyse the influence of different exogenous and endogenous factors on the fore‐ cast error of the movie exchange. Exogenous factors are the ones which are not derived from the virtual stock market itself, but rather depend on the product being used on the virtual stock market: the distribution intensity of movies in the form of the number of screens a movie is released on opening weekend as well as the genre of a movie (see Table 4). Endogenous to the stock market is the stock price volatility of a specific stock on the last 5 days of trading at the virtual stock market.
Table 4:
Movie Genre
Coding According to Genre of Movie
Action/Thriller 22 Drama/Romance 18 Comedy 26 Rest 15
Number of Movies
ANOVA (Impact of genre on forecast error): F-Value = 1.116, p-value = .348
12
Opportunities of Virtual Stock Markets to Support New Product Development
Table 5 displays the estimation results for the influence of endogenous and exogenous factors on the forecast error of all 81 movies traded at the movie exchange. Thereby, only price volatility and the number of movie screens at the opening weekend exert a significant influence. However, these two variables (one exogenous and one endoge‐ nous) display a significant negative correlation (Pearson: ‐.400 (p‐value: .000)). Thus, both, the exogenous factor of the number of screens, as well as the endogenous factor of price volatility can indicate the expected forecast accuracy of the virtual stock mar‐ ket. If we omit from the 81 movies the 20% having the highest price volatility, then the mean forecast error reduces from 211.9% to 97.13%. The cut of value for the price vola‐ tility is a coefficient of variation of 0.509 or 50.9% in this case. Analogously, if we omit from the 81 movies the 20% having the lowest number of screens on opening weekend, then the mean forecast error reduces from 211.9% to 69.79%. The cut of value for the number of screens is 119 in this case. The latter results are in line with the forecasting errors (71.1%) of Sawhney and Eliashberg (1996) in a study to predict the box‐office revenues for ten movies.
Table 5:
Estimation Results for Factors influencing Forecast Error of Movie Exchange
Model 1 (.008)
b
Parameter value (standardized) Constant (p-value)
a
Model 2 (.000)
Model 3 (.613) .277 (.012)
Price volatility (p-value)
.153 (.199) -.289 (.024) -.252 (.078) -.227 (.123) -.229 (.127) .171 3.084 (.014) .102 8.955 (.004) -.319 (.004)
Number of Screens (p-value) DV_Action_Thriller (p-value) DV_Drama_Romance (p-value) DV_Comedy (p-value) R
2
.077 6.584 (.012)
F-Value (p-value) N = 81 Movies
a b
Constant: No value for standardized parameters. Measured as coefficient of variation.
5
Summary and Conclusions
The results of the empirical study show that virtual stock markets can provide better predictions than expert judgments. Yet, there is no guarantee that virtual stock mar‐
13
Bernd Skiera, Martin Spann
kets lead to good results and the empirical study shows that virtual stock markets might also provide some rather weak forecasts. The promising result, however, is that the forecasting error might be further reduced by recognizing the factors that had a negative influence on forecasting accuracy in previous virtual stock markets. There‐ fore, the repeated use of virtual stock markets allows to develop good indicators for the expected forecast accuracy and the price volatility might serve as a general indica‐ tor for a VSMʹs predictive validity. Virtual stock markets seem to provide promising opportunities to support the new product development process and the recent publications indicate that the use of vir‐ tual stock markets might provide many benefits for companies. As most of these op‐ portunities have gained very little attention in literature, virtual stock markets might be a rich field for further studies in the area of new product development. The avail‐ ability of a flexible software solution (www.virtualstockmarkets.com) will support research in this area.
6
References
Albers, Sönke, Claudius Paul, and Matthias Runte (1998): Virtuelle Communities als Mittel des Absatzes. Otto Beisheim (ed.): Distribution im Aufbruch. Bestandsaufnahmen und Perspektiven, Munich: Vahlen Verlag. Brockhoff, Klaus (2000): Produktinnovation. Handbuch Produktmanagement, Sönke Albers and Andreas Herrmann, Eds. 2 ed. Wiesbaden: Gabler Verlag. Brockhoff, Klaus (1999): Produktpolitik, 4th ed., Stuttgart et al.: Gustav Fischer. Chan, Nicholas T., Ely Dahan, Adlar Kim, Andrew W. Lo, and Tomaso Poggio (2002): Securities Trading of Concepts (STOC). Cambridge, MA: Massachusetts Institute of Tech‐ nology. Dahan, Ely and John R. Hauser (2002): The Virtual Customer, Journal of Product Innova‐ tion Management, 19 (5), 332‐53. Dahan, Ely and V. Seenu Srinivasan (2000): The Predictive Power of Internet‐Based Product Concept Testing Using Visual Depiction and Animation, Journal of Product Innovation Management, 17 (3), 99‐109. Di Benedetto, Anthony C. (1999): Identifying the Key Success Factors in New Product Launch, Journal of Product Innovation Management, 16 (6), 530‐44.
14
Opportunities of Virtual Stock Markets to Support New Product Development
Eimeren, Birgit, Heinz Gerhard, and Beate Frees (2003): ARD/ZDF‐Online‐Studie 2003. Internetverbreitung in Deutschland: Unerwartet hoher Zuwachs, Media Perspektiven (8), 338‐58. Eliashberg, Jehoshua, Jedid‐Jah Jonker, Mohanbir S. Sawhney, and Berend Wierenga (2000): MOVIEMOD: An Implementable Decision‐Support System for Prerelease Mar‐ ket Evaluation of Motion Pictures, Marketing Science, 19 (3), 226‐43. Ernst, Holger, Jan Henrik Soll, and Martin Spann (2004): Möglichkeiten der Lead‐User‐ Identifikation in Online‐Medien. Cornelius Herstatt and Jan Sander (ed.): Produktent‐ wicklung mit virtuellen Communities, Wiesbaden: Verlag Gabler. Ernst, O. and Henrik Sattler (2000): Validität multimedialer Conjoint‐Analysen. Ein empirischer Vergleich alternativer Produktpräsentationsformen, Marketing ZFP, 22 (2), 161‐72. European Information Technology Observatory (2004): 10th Edition. Frankfurt am Main. Forsythe, Robert, Forrest Nelson, George R. Neumann, and Jack Wright (1992): Anat‐ omy of an Experimental Political Stock Market, American Economic Review, 82 (5), 1142‐ 61. Forsythe, Robert, Thomas A. Rietz, and Thomas W. Ross (1999): Wishes, Expectations and Actions: A Survey on Price Formation in Election Stock Markets, Journal of Eco‐ nomic Behavior & Organization, 39, 83‐110. Grover, Varun and Thomas H. Davenport (2001): General Perspectives on Knowledge Management: Fostering a Research Agenda, Journal of Management Information Systems, 18 (1), 5‐21. Hayek, Friedrich August von (1945): The Use of Knowledge in Society, American Eco‐ nomic Review, 35 (4), 519‐30. Hulse, C. (2003): Pentagon Prepares a Futures Market on Terror Attacks, The New York Times, 29. July 2003. Ortner, Gerhard (2000): Aktienmärkte als Industrielles Vorhersagemodell, Zeitschrift für Betriebswirtschaft ‐ Ergänzungsheft, 70 (1), 115‐25. Panten, Gregor, Claudius Paul, and Matthias Runte (2001): Virtuelle Communities. Sönke Albers and Michel Clement and Kay Peters and B. Skiera (eds.): Marketing mit Interaktiven Medien. Strategien zum Markterfolg, 3rd ed. Frankfurt am Main: F.A.Z.‐ INSTITUT. Plott, Charles R. (2000): Markets as Information Gathering Tools, Southern Economic Journal, 67 (1), 1‐15.
15
Bernd Skiera, Martin Spann
Polk, Charles, Robin Hanson, John Ledyard, and Takashi Ishikida (2003): The Policy Analysis Market: An Electronic Commerce Application of a Combinatorial Information Mar‐ ket, in ACM Conference on Electronic Commerce. San Diego. Rayport, Jeffrey and Bernard J. Jaworski (2001): E‐Commerce. New York: McGraw‐Hill Higher Education. Sawhney, Mohanbir S. and Jehoshua Eliashberg (1996): A Parsimonious Model for Forecasting Gross Box‐Office Revenues of Motion Pictures, Marketing Science, 15 (2), 113‐31. Smith, Vernon L. (1982): Microeconomic Systems as an Experimental Science, American Economic Review, 72 (5), 923‐55. Spann, Martin (2002): Virtuelle Börsen als Instrument zur Marktforschung. Wiesbaden: Gabler Verlag. Spann, Martin, Holger Ernst, Bernd Skiera, and Jan Henrik Soll (2003): Identification of Lead Users via Virtual Stock Markets. Frankfurt am Main: Johann Wolfgang Goethe‐ Universität. Spann, Martin and Bernd Skiera (2003a): Einsatzmöglichkeiten virtueller Börsen in der Marktforschung, Zeitschrift für Betriebswirtschaft, zur Veröffentlichung angenommen. Spann, Martin and Bernd Skiera (2003b): Internet‐Based Virtual Stock Markets for Business Forecasting, Management Science, 49 (10), 1310‐26. Spann, Martin and Bernd Skiera (2003c): Taking Stock of Virtual Markets. How can Internet‐Based Virtual Stock Markets be Applied for Business Forecasting and Other Forecasting Issues, ORMS Today, 30 (5), 20‐24. Toubia, Olivier, Duncan I. Simester, John R. Hauser, and Ely Dahan (2003): Fast Poly‐ hedral Adaptive Conjoint Estimation, Marketing Science, 22 (3), 273‐303. Urban, Glen and Eric Von Hippel (1988): Lead User Analyses for the Development of New Industrial Products, Management Science, 34 (5), 569‐862. Urban, Glen L. and John R. Hauser (1993): Design and Marketing of New Products (1 ed.). Wertenbroch, Klaus and Bernd Skiera (2002): Measuring Consumer Willingness to Pay at the Point of Purchase, Journal of Marketing Research, 39 (May), 228‐41. Wolfers, Justin and Eric Zitzewitz (2004): Prediction Markets, Journal of Economic Per‐ spectives, forthcoming.
16