Two essays on a sales forecasting model incorporating replacement demand and an adaptive forecasting model considering structural changes

Abstract

In the product life cycle, mature stage is defined as the period of time when products have penetrated into almost all of potential customers and their sales volume reaches a steady state. For some products the maturity phase can be the longest stage as the products are repeatedly purchased by loyal customers. However, while overall sales may grow year-over-year, percentage sales increases may be small. In order to forecast sales of mature market products, Box-Jenkins time series analysis has been used for short-term forecasting, and regression analysis considering many demand factors has been used for long-term forecasting. Since these general forecasting methods do not consider specific characteristics of mature market, they must have some limitations. This thesis suggests tailor-made approaches for forecasting sales of mature market products. They are divided into two main approaches. The first approach is the long-term forecasting model for aggregate level mature market products (A choice-based multi-product (brand) diffusion model incorporating replacement demand) and Short-term forecasting model for individual level mature market products (Bayesian adaptive exponential smoothing). With proposed approaches, we could obtain better forecasts for mature product sales, and deeper understanding of customers in mature market.