Previous Section
Now, the problem is that I've got a lot of parameters in this model, right? I've got this entire cross-price elasticity matrix, I've got all these feature and deal effects, so I can't just go out and say there are 192 different demographic relationships. This is going to be another problem I've got. So, I know this is little hard to read, but essentially this is my model . The movement from one particular brand is equal to the cost, times all these price effects, times these feature effects, times these promotional in-store deals, and the lag effects. There are some economic motivations and there's reason to expect that the demographic effects within each of the prize-quality tiers of these are going to be similar. So if education is positively affecting the price sensitivity of premium orange juices, it's going to affect not only Tropicana but also if Minute Maid has a premium orange juice. I'm also going to say that there is some type of common effect where the only price elasticity is of the national brands. Now, I didn't have 12 colors, but if I did I would have colored each of these squares differently. The point is that just because they're green, it doesn't mean that all of these green stores are going to have the same effects. There's going to be one effects, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, thirteen, fourteen, fifteen. There are 15 sets of these demographic relationships between my price elasticities and my feature coefficients and my demographics.
Go to written version of paper