Care must be taken when evaluating those statistics because all the ones I've seen are done only in dollar terms and entirely discount the economic beneficiarie

Care must be taken when evaluating those statistics because all the ones I've seen are done only in dollar terms and entirely discount the economic beneficiaries of the movement of people and goods. For example, consider a paved rural road into farmland. In dollar terms paving that road is a subsidy from the nearest city to the people living in that rural area. However, in part the road is paved rather than gravel only to support heavier trucks to more efficiently transport agricultural products destined for the city. It is also paved in part to support larger, faster, heavier agricultural equipment which brings economies of scale to agriculture and reduces the per-unit price of the result -- again destined mostly for the city. Residents themselves don't need the more expensive paved and it isn't their relatively light private vehicles causing most of the wear on the road in the first place. Considered this way, a not insubstantial fraction of the cost of non-city areas is the city indirectly subsidizing itself. The full costs could be incorporated directly into the goods sourced from the supposedly subsidized areas, but that would be less efficient overall. For example, good roads reduces the cost of agricultural products for, say, three months a year. Instead of directly paying capital costs to pave the rural road the city could pay operational costs in higher food prices while missing the other cost advantages of the paved road the rest of the year in reduced recreational costs, policing costs, education costs, etc. If those other costs were higher there would be less of them and people would be less willing to live in those areas, increasing wage and commuting costs.