Consider n firms (agents) located at a river, indexed by \(1, \dots , n\) from upstream to downstream. The pollution generated by these firms induce cleaning costs \(c_1, \dots , c_n\), where \(c_i\) is the cost for cleaning the water in region i (according to the local environmental standards).
The corresponding cost allocation problem is highly interesting both in theory and practice.
Among the most prominent allocation schemes are the so-called Local Responsibility and Upstream Equal Sharing. The first one allocates simply each local cost \(c_i\) to the corresponding firm i. The second distributes each \(c_i\) equally among firms \(1, \dots , i\). We propose and characterize a dynamic scheme which, given a particular order of arrival, allocates the current total cost among the firms that have arrived so far. The corresponding expected allocation (w.r.t. a random arrival order) turns out to be a convex combination of the two schemes above.