The group discounted to a large extent the theoretical aspects of the paper. The general line and thoroughness of the analysis was endorsed as relevant and practical. Correspondingly the distinction between descriptives versus function status of the model was ignored on the policy of "don't worry about the validity of an 'explanation' until you've got one". Put another way, the group viewed the paper as a psychologically informed way of collecting and reviewing data and not as an explicit psycho-sociological model of behaviour.
The method used to obtain optimal or quasi-optimal hierarchy of the activities correlated with the propensity to purchase was not mentioned. Some people insisted on the existence of quite different methods, either deterministic or probabilistic, to perform such analyses. We asked the speaker to give more details on this part of his work.
The discussion of the contributions of B. Baldrey and K.S. Feldman and J.F. Boss consisted at first of the classification of both systems.
This model is purely a financial one. To give the inputs the different departments of the company must have their own models. The advantage of the corporate financial model is to make the different departments aware of the necessity of computing the inputs and to make the lack of information appear.
On the first place, it has to be said that the group members consider it as a jolly good idea to observe the purchasing habits over quite long periods of time instead of making conclusions based on elementary purchasing behaviour. But they regret that the authors have not (or not yet) studied the relationship between the duplications and the D coefficients observed for specified pairs of brands during successive periods. It could be an interesting direction for research to analyse these data on the way it has been done with elementary purchases when applying Markovian processes.
The two approaches had been introduced as alternatives. As such, the group agreed with Ehrenberg & Goodhardt's approach. They did, so on purely practical grounds . The crucial drawback seen in Hamre's approach was that the number of constants or coefficients to be fitted from the data kept increasing. The group felt that this would lead to instability under sampling variation.
We have mainly discussed the problem of weighting properly the variables. It is quite clear that the authors do not want to introduce any discriminating policy between the items of variables. But they consider it as a necessary refinement to associate with each of these items a weight expressing its discrimination power. Their similarity index is therefore weighting the variables according to their information content.
One question which has been raised in the discussion is that of assessing a manufacturer's scope for launching an additional brand in a product-field instead of simply pushing his existing brand. This is in fact one of several kinds of practical application of the results which are already being actively pursued. It links up with the more general topic of studying buyer behaviour for aggregates of brands. This has specific applications not only for judging the scope for additional upper sales limits of existing brands and for assessing the role of house-names, but also provides a link with macro-economics.
The number of people who buy two particular brands in a given time period does not depend on the brands as such but only on the numbers of people who buy each brand and on a general constant. The constant varies by product-field and by the length of the analysis-period, the correlation between buying any two brands being positive in relatively long periods and negative for shorter periods.
The factors entering into the financial planning of a large corporation are manifestly intricate, and even in the few instances in which the long term objectives are unambiguously defined it is rarely easy to see how these can be attained. In some cases it is by no means obvious what the objectives should be, but it is hoped that at the corporate level our model could be used to examine the financial implications of alternative decisions and thereby assist in the setting of targets. The development of on-line computer facilities at a reasonable cost allows programs (such as ours) to perform in conversational mode with the user taking decisions during the run. The usefulness of any completely automatic simulation model is limited by the difficulties of quantifying the company's objectives. These are best summed up in somewhat vague financial and socio-political terms. The model has little to offer in the latter field, but when operated by skilled top management should have a significant contribution to make to the former. It would, for instance, be difficult to construct an algebraic formula which draws attention to the possibility of being taken over, but a financial expert using the model would be able to detect the danger signals from the business ratios, balance sheet and profit and loss account items printed out. This would, in turn, help him to isolate the investment decisions that led to this situation.