A company exports a units of product to a distributor in a foreign country. This distributor then sends units to dealers across the country who run shops that sell the product.
Until May 2017, distributor was given money each year earmarked for branding their dealers. Branding is meant to increase sales. After May 2017, branding was transitioned to the company itself, and the distributor gets no money to do so.
I have data on the number of units sold by each dealer every month from start of 2016 to March 2018. I am to find whether branding expenditure done by the company correlates with sales growth. If it does, I want to find how much growth is attributable to that expenditure.
I was considering this approach: Create a time-varying indicator for branding coded to 0 prior to the transition and 1 after the transition; use this to forecast sales for 2017; take the difference between expected and actual sales from May 2017 onward and call this amount the contribution of dealer branding.
- Is this a correct approach?
- Are there assumptions I should be considering?
- What other approaches might you suggest?