I am working on a panel dataset on public procurement. Each row represents a company bidding in a specific tender and submitting a bid. I am performing a regression analysis to predict when a company is colluding (I have this information from a judicial court document). I the regression, among other variables, I would like to test as predictors: a) number of bids submitted by each company b) number of times companies bid in temporary associations
From a theorethic point of view it makes sense, as a) explains whether aggressive bidding may predict collusion, and bidding in temporary associations is an indicator of the phenomenon as accounted by competition authorities.
The problem (I don't know if it is) is that b) is derived from a). a) comprises ALL the offers submitted by a company, while b) just those presented jointly with other companies.
Is this a problem for regression? What do you suggest? Is it maybe better to insert in the model b) and eventually c) (number of offers NOT in temporary associations)?
Hope to be clear.