Competition in the Retail Market of Consumer Packaged Goods Defended on Thursday, 14 January 2021
This dissertation investigates competition in the retail market of consumer-packaged goods from different angles. Chapter 2 studies how brands react to each other’s price promotions, the focus is put on the asymmetric reactions between brands with different market shares and price levels. To this end a Hierarchical Bayes Ordered Probit model (HB-OP) is employed to study the moderating factors on reactions. The results show that the reaction intentions are influenced by brands’ relative market shares, together with some category specific characteristics. Chapter 3 investigates the competition between retail chains and the role of their private label brands. We propose a Hierarchical Bayes Market Share (HB-MS) model to investigate how a retailer’s market share depends on price changes by national brands and private labels, and how the baseline market share and the price sensitivities are influenced by private-label positioning. Chapter 4 aims to compare traditional sales forecasting models with modern techniques like factor models, Lasso, elastic net, random forests and boosting methods. We consider all possible brands as potential competitors that might be useful for the sales forecasts of a focal brand. This approach is relevant if we do not know beforehand which brands have predictive content, and in this case, we can let the data help to decide on this each time we make a forecast. The forecasting accuracy of a variety of models are compared across a large number of brands
Keywords
Price promotion, competition, retail market, private-label, market-share, forecast, Bayesian, machine learning