Long term sales forecasts of innovations - an empirical study of the consumer electronic market

Entrepreneurial action, whether undertaken by firms small or large, new or old, involves the ability to recognise and exploit opportunities. This is a risky activity. The consumer electronic industry (the focus of this study) is characterized by high development and launch costs of technological innovations as well as high failure rates. Moreover, technological innovation tends to be an ongoing pursuit of consumer electronic companies. Accurate long term forecasting of sales, and understanding the drivers of those sales, assists managers to better recognise and exploit opportunities over the life cycle of a product category. In durable product categories such as consumer electronics, accurate long term forecasting models need to incorporate first and repeat purchases. This is important both because the latter represents a significant share of sales for durables over time, and because consumer requirements for product innovation develops as they become experienced, sophisticated users of the product. While traditional aggregate diffusion models for durables mostly consider first purchases only (e.g., Bass 1969 and its many extensions), less attention has been devoted to diffusion models considering repeat purchases. This is due to the fact that this research tradition has largely relied on secondary data which do not offer a distinction between first and repeat purchases. This makes it very difficult to assess the validity of any model incorporating repeat purchase. Therefore,our study undertakes a large primary data collection to address this issue. In the diffusion literature, there is no consensus regarding which repeat purchase model is most suitable for forecasting purposes. Most of these models are developed on the basis of aggregate annual sales data which do not provide any information on the individual buying behaviour. One weakness of the repeat purchase models published so far is the failure to measure heterogeneity of consumer's repeat purchasing behaviour. Thus, modelling this component adequately might lead to a better overall forecasting performance. For this reason, individual consumer data, rather than aggregate sales data, was collected in this study.