Drissel on the Normative Core
By Bryan Caplan
I received the following email from Bill Drissel about my “Public Choice: The Normative Core.” Reprinted with his permission.
The data you seek for your “normative core” is readily available in one arena: public transportation. I follow the Anti-Planner, Randal O’Toole. The planned benefit is number of riders. The planned cost is usually available in dollars(of a given vintage). The subsequent cost-overruns and consequent ridership are also available. So every cost/benefit ratio could easily be adjusted by a “normative core adjustment”.
For example, if extensive research shows that, for public transportation, actual costs / planned costs average 2.1 and actual ridership / planned ridership average 0.40, then the normative core adjustment factor for public transportation = 5.25. So for back-envelope calculations, the planned cost/benefit should be multiplied by 5 or so.
I have considerable acquaintance with software estimates including many that ended up in proposals and contracts. I don’t know of a single case of software under-run (costs less than planned). I would guess the typical over-run for routine software development at 2:1. For difficult stuff: 5:1. Really hard stuff like voice, face, fingerprint recognition: much higher than 5. Development of a capable word processor with fonts and embedded images like MS Word, a single lifetime wouldn’t be enough.
I had a one-man consulting business for 45 years. Whenever I asked a client about his over-run experience, I got a mournful story. If I suggested he apply an experience-based multiplier, the response was always, “If we did that, we’d never get any business!” I guess that’s the equivalent of, “Junior Professor: By that standard, government should never do anything.”
I admire the work that you and Don Boudreaux do. I’m 87 but if I had encountered the public choice body of knowledge while I was much younger, I might have given up engineering for economics.
The Colony, TX