Date of Award

Spring 1999

Document Type

Dissertation

Degree Name

Doctor of Business Administration (DBA)

Department

Marketing and Analysis

First Advisor

Hani Mesak

Abstract

This study, using the dynamic programming approach, has addressed the problem of optimally allocating a fixed advertising budget of a monopolistic firm over a planning horizon comprised of n equal periods to maximize two popular measures of advertising performance: (1) profits related to the advertising effort (discount factor r = 0), and (2) present value of profits related to the advertising effort (discount factor r > 0).

Two dynamic programming models that use the modified Vidale-Wolfe model to represent sales response to advertising are formulated with respect to whether the time value of money is considered. For a planning horizon comprised of four equal time periods, computing routines are developed to solve two sample problems with respect to the dynamic programming models. Sensitivity analyses are performed to assess the impacts of a change in some key model parameters upon the behavior patterns of the optimum dynamic programming advertising policy and the associated total return.

Four alternative types of traditional advertising pulsation policies are modeled for the purpose of comparing their performance with the optimum advertising policy determined by dynamic programming. For a planning horizon comprised of four equal time periods, computing routines are also developed to generate total returns under these traditional advertising pulsation policies. Computational results show that the performance under the optimal advertising policy determined by dynamic programming, as expected, is at least as good as the maximum performance among the four traditional advertising pulsation policies.

The plausibility of the modified Vidale-Wolfe model is empirically examined using the well-known Lydia Pinkham vegetable compound annual data covering the period from 1907 to 1960. Model parameters have been estimated using the Gauss-Newton algorithm related to nonlinear regression. The model selected is one corrected for first-order autoregressive residuals. The empirical results indicate that the model parameters are statistically significant and of the expected signs. More important, it is found that the advertising response function is concave.

Included in

Marketing Commons

Share

COinS