| Posted in DVD | Posted on
To develop a behaviourally-motivated model of aggregate DVD preorder/sales patterns.
Design/methodology/approach - Models the purchase timing decision of an individual consumer using an optimal stopping framework that explicitly captures forward-looking behaviour. Allows consumer utilities to be heterogeneous and derives the aggregate preorder/sales curve and then extends the model to handle a segment of consumers who remember the release date.
Findings - The DVD-II model performs better on all measures than the DVD-I model, which in turn outperforms state-of-the-art benchmark models, such as the Moe and Fader (2002) model and the Weibull-Gamma model, adding that the DVD-II model obtains the highest log-likelihood, the lowest MAPE and the lowest MRAE in comparison with all models. Notes that it has a lower BIC than DVD-I, indicating that the two-segment model captures the actual data more closely than the single-segment model, and its overall superiority is further revealed by boxplots of the RAE for each DVD across the four models. Believes that the models perform better than the benchmark models because the behavioural premise behind them is more consistent with the DVD context.
Research limitations/implications - States that the model can be further extended by incorporating non-stationarity, including dynamic pricing, investigating movie characteristics and undertaking cross-category analysis and joint modelling of box office and DVD revenue.
Practical implications - Contends that the model can generate insights through policy experiment, estimation of residual sales and forecasting post-release sales based only on preorders.
Originality/value - Explains the temporal preorder/sales pattern of DVDs using an individual-level modelling framework.
Keywords: CONSUMER BEHAVIOUR, DVD, SELLING, Article Type: Research paper
Design/methodology/approach - Models the purchase timing decision of an individual consumer using an optimal stopping framework that explicitly captures forward-looking behaviour. Allows consumer utilities to be heterogeneous and derives the aggregate preorder/sales curve and then extends the model to handle a segment of consumers who remember the release date.
Findings - The DVD-II model performs better on all measures than the DVD-I model, which in turn outperforms state-of-the-art benchmark models, such as the Moe and Fader (2002) model and the Weibull-Gamma model, adding that the DVD-II model obtains the highest log-likelihood, the lowest MAPE and the lowest MRAE in comparison with all models. Notes that it has a lower BIC than DVD-I, indicating that the two-segment model captures the actual data more closely than the single-segment model, and its overall superiority is further revealed by boxplots of the RAE for each DVD across the four models. Believes that the models perform better than the benchmark models because the behavioural premise behind them is more consistent with the DVD context.
Research limitations/implications - States that the model can be further extended by incorporating non-stationarity, including dynamic pricing, investigating movie characteristics and undertaking cross-category analysis and joint modelling of box office and DVD revenue.
Practical implications - Contends that the model can generate insights through policy experiment, estimation of residual sales and forecasting post-release sales based only on preorders.
Originality/value - Explains the temporal preorder/sales pattern of DVDs using an individual-level modelling framework.
Keywords: CONSUMER BEHAVIOUR, DVD, SELLING, Article Type: Research paper
