The diffusion of new consumer durables and the role of advertising
Existing economics literature, especially empirical, on the role of advertising in the diffusion process is sparse. In this thesis, explicit attention was given to the part played by advertising in the diffusion of new consumer durables, both at a theoretical and empirical level. The two competing epidemic and probit demand diffusion models were used and two period models established. Advertising was incorporated into each framework, becoming endogenous once a supply side was added. The resulting profit maximising advertising sales ratio within each model was compared and contrasted to the standard Dorfman-Steiner and Nerlove-Arrow conditions found in the literature. The special features of the models did impinge upon these ratios. Alternative supply structures (defined in terms of number of firms) were considered and the resulting effects on the industry advertising sales ratio within each demand framework explored. The results could not be rigorously shown. however the essence of each of the demand frameworks had potentially different implications for the monopoly as compared to oligopoly outcome. The existence of a positive generic information externality in the epidemic model, reducing the advertising intensity once the monopoly assumption was relaxed, whilst the negative early extraction effect and intertemporal price discrimination opportunities in the probit model could increase the advertising sales ratio in an oligopoly industry. Two new consumer durables were chosen for the empirical investigation, the video cassette recorder and the colour television receiver. Both demand diffusion models were applied to the data for each product using OLS estimation procedures. It was found that the epidemic model performed relatively better to the probit model for the video cassette recorder, indicating a positive and statistically significant role for advertising, when entering indirectly through the social contact coefficient. Thus the information variables took precedence for this particular product. However there was some question over the predictive ability of the model and no account was taken of any possible simultaneous equation bias that would arise if advertising was endogenous, as suggested in the theoretical chapters. In contrast, the probit model performed relatively better for the colour tv receiver, showing the importance of the economic variables such as relative price, hire purchase restrictions and total advertising messages, in addition to the existing owners to non owners explanatory variable. This conclusion was reached after recognition of the possible existence of simultaneous equation bias. The information about the advertising decision contained in the theoretical chapters was used to identify the instrumental variables to include in the TSLS estimation procedure. However, restricting the sample period to the first 11 years of the diffusion process cast some doubt upon the probit model specification. This might indicate that future work should consider incorporating information variables into the probit structure in a more satisfying manner.