Tuesday, July 3, 2012

That moral hazard thing

Moral hazard: we learn these from the knowledge base at wikipedia:
A moral hazard is a situation where there is a tendency [by a work package manager, for example] to take undue risks because the costs are not borne by the party taking the risk.

And, somewhat related, we are informed that "adverse selection" is described this way:
Adverse selection is a situation where an individual's demand for [relief from risk] is positively correlated with the individual's risk of loss (e.g. higher risks buy more insurance), and the [project manager] is unable to allow for this correlation in the [project plan]

This discussion about moral hazard came up with some of my students in my Agile PM class. We were discussing how tightly to schedule a project.

My thinking: a plan without slack is not really a plan; it's a hope.

Two ways to get slack planned into the project baseline:
  1. Assume there's always going to be "labor loss"... That is, no one can really work 100% of the time they are schedule to work. It just doesn't ever seem to work that way. From dental appointments to coffee bar chat, some labor is lost. I always estimate 15%
  2. Put in what I call "pipeline buffers", to allow some "breathing" in the schedule plan. (I may call it pipeline buffers, but others call it critical chain method) 

(Many reading this blog may have never seen an install of pipelines; nevertheless, I'll describe it by saying that what they do is put in expansion joints every few lengths by letting the pipe zig and then zag. These zig zags absorb the changing length of the pipe with temperature, something like an accordion, and something you have to worry about in the physical world)

So, now moral hazard: the team knows there is slack built-in. In the agile business, we don't schedule for more than 85% of velocity, and we always put in a zero activity iteration before every release (a release is some number of iteration's deliverables). If the team knows this, it creates a moral hazard. They know they can use the slack without having to pay the price of the longer schedule.

And, that creates the correlation of the adverse selection dilemma: a demand for schedule relief on the one hand, and a correlation with the propensity of work packages and iterations to run over.

What's a PM to do to fight moral hazard? Here are a few ideas:
  • Incentives to do good, i.e. pay for performance (or 'show me the money')
  • Penalties if the schedule is blown
  • Keep the labor loss to yourself; that just happens. No point in offering it up to the general population
  • Make it hard to use the buffer time (after all, the PM should be in charge of constraints, not the inmates)
In effect, any schedule that's going to work has got some slack (some room to zig and zag)