You have a 30-yearpanel dataset on a large random sample of firms in Europe of different sizes.Assume that you have data on all reasonable controls that you can think of andthat there is no selection bias in your sample. You are asked to study whethersales affects the profits of the firm per unit of sales and whether profits perunit of sales affects sales. How would you study and test these tworelationships?


大家有什么高见？