Large scale cost/revenue analysis


I'm currently developing a tool to analyze a vast amount of information for cities across Canada. I cam across your lafayette example a month or so ago.

We developed a methodology to calculate residential property tax revenue per square km for the biggest metropolitan area (which represents between 50 and 80% of city's revenue). We also did some commercial property tax revenue per square km. We havent figured out how to split the other revenues (mostly transfers from other level of governments).

I'm mostly struggling with the expense side of the equations. I looked at municipal financial statements and budget and couldn't find a ventilation at the dissemination area or dissemination bloc level. Then i thought of using a simple formula that would look like this: split budget between the DA or DB based on % of population, % of area, % of km of roads, and % of dwelling; and then do a straight or weighted average of these 4 elements. I thought about adding other metrics such as median income, transit access, density, etc, but I’m not sure the level of precision would be way better. After consulting with researchers and experts we dropped this approach.  

in your article i noticed that you interviewed city employees. unfortunately, this is not something that is realistic due to the number of cities included in our analysis. I was curious to get your views on how this type of work could be done at the national level. Thanks a lot.


1 comment

  • Comment author
    Charles Marohn

    Yeah, you've reach the same problem I reached back in 2007. You can't scale costs this way, not in any reliable fashion. Even here in central Minnesota, with 200 feet of sandy glacial outwash to work on, the cost of a road will vary greatly from project to project based on soil conditions alone. Now add the infinite other variables and there is really no way to standardize, and underground work is even more complicated.

    In Lafayette, we took four trips and spent hundreds of hours trying to understand their ongoing costs. Even then, with loads of hard data, there was still credible pushback that we hadn't gotten things correct. I'm comfortable we were really close -- and the margins between their revenue and expenses were so massive that getting a higher degree of confidence provided no real value.

    So, I hate to disappoint, but I don't think this work can be credibly done on a national level. The people I've seen try end  up producing stuff that reads more like propaganda than serious research (due to the assumptions). This is why I pretty quickly started to focus on micro-examples (project by project) and used those insights to construct a larger set of insights about how these complex systems work. My assumption was that, if the individual projects I could model with confidence were way out of whack financially, and the system as a whole was way out of whack financially, that I didn't need to know the replacement cost of ever pipe or road to feel pretty confident in the connection.

    Add in extensive study and research into human psychology and group decision-making, and you have the methodology that produced the core insights of Strong Towns.


Please sign in to leave a comment.