Previous Section
If we look at these effects in terms of this gamma matrix the gamma vectors that I'm estimating -- .I've got 15 of these gamma vectors because I've estimated for a lot of different old price sensitivities, for the national brands, for the premium brands, for the store brands. Now I'm interested in thinking about -- well, let's go across the matrix instead of going down it. Let's think about the effects of the percentages of households with a value of 150 thousand or more on price response. The one we just went through was the national owned, so before I show you, in this case the more households with values of 150 thousand or more, the less price sensitive the store is.
So let's think about this -- in this case we're getting less price sensitivity as the wealth effect increases, or the more houses you get over 150 thousand. Let's think about some of these other parameters. So for the premium -- you see that this effect is even greater. For the store brands, this effect is smaller. In fact, it's going in the other direction, although it's difficult to say this is getting close to ground zero. Then you can start looking at the effect of larger household sizes, or larger household values, the substitution between the brands. Sometimes its going to be less, so the more households you have over 150 thousand, the less substitution you'll have between premium brands. Before I did that, it wasn't exactly clear to me. I had some notion of what the old price elasticities would be (that the wealth effects are going to drop it), but it's a lot more difficult to tell the story of what's going on with substitution between brands. I think this gives some really interesting results.
To see the relationship between other coefficients (store, premium, own-price, cross-price and feature), click here.
To see parameter estimates for the cross price elasticity matrix click here.
Go to written version of paper